October 2026 status archive, page 1 of 2
Shutdown notices, recent changes and all monthly archives
These events link to today's eligible findings, not historical before-and-after claims.
Cited events
1 ADE
Vibe Kanbaninitial researchresearch correctionfreshness recheck43 findings · 33 sources
Findings today
- The continuing local product provides browser workspaces, a product CLI and Tauri desktop packages.
- Released source integrates multiple coding-agent adapters; actual mixed-harness execution was not tested.
- Local Codex uses native login or API keys and detects custom configuration directories.
- The documented local flow is npx, agent authentication and repository selection.
- Local MCP provides workspace start, session creation/follow-up and execution inspection.
- Vibe Kanban manages per-harness MCP configuration; connections and permissions execute through the selected harness.
- The desktop distribution uses Tauri webviews and a web UI.
- Local prompt-driven workspaces continue independently of the retired hosted issue tracker.
- Local plan approval and workspace notes coexist with explicit archive retention.
- Local sessions share files and reusable prompts while keeping conversations separate.
- Local chat accepts file mentions and images; architecture-aware retrieval and enforcement remain untested.
- The vendor-hosted issue board is retired.
- Docs describe independent sessions that can run simultaneously and multiple workspaces.
- The pinned release ships an explicit orchestrator MCP mode.
- In the pinned local continuation, work starts through explicit session/workspace requests.
- Local chat provides stop, queue/retry and history controls.
- The local workspace list supplies session status context, but a 12-state fleet/epic fixture was not evaluated.
- Local users can try a different approach by duplicating a workspace, which keeps the repository/branch configuration…
- Local agent execution continues.
- Branches/worktrees separate workspace files; sessions within one workspace share files/Git.
- Local archive preserves history/files and missing worktrees rebuild from last commit with uncommitted changes lost.
- Repository setup/dev/cleanup scripts prepare environments with user-defined lifecycle behavior.
- Approval timeouts are documented; enforced time, token and spend budgets remain unverified.
- Local logs include expandable tool command/search results and execution states.
- The documented context gauge is not a verified cost ledger.
- Resource limits, check reuse and fleet backpressure remain unknown.
- The shutdown announcement retired vendor-hosted issue discussion/comments.
- The vendor-hosted organisation/project service was named for retirement.
- Harness-specific plan/permission controls appear in local configuration.
- Local conversation history persists across session switches and archive, but message retry can replace later visible…
- Local review UI provides diff modes, filtered files and submitted GitHub comments.
- Local archive retains conversation notes/files, while diffs and previews remain separate review views.
- Repository scripts can run checks; docs advise review and CI before merge.
- Local workspace docs cover branches/worktrees, PR creation, push, merge and rebase/conflict handling.
- Local merge is documented.
- Git PR workflow is documented, but completion-to-revision evidence, rollback and stale-approval invalidation were…
- Single-workspace review is documented; six-output batch review, dependency queues and exact-revision approvals…
- Pinned repository distribution has an Apache-2.0 license.
- Local workspaces continue without vendor sign-in.
- Local telemetry has a settings toggle and a released source default of enabled, but a successful sign-in turns…
- Commercial services were sunset while the local open-source project continues.
- The local open-source project remains available without vendor sign-in.
- Commercial subscriptions were terminated, so historical Vibe Kanban tariffs do not describe a current service.
Sources
Open the Vibe Kanban page
11 ADEs
Google Antigravityfreshness recheckresearch correction41 findings · 32 sources
Findings today
- Desktop on macOS 12+ with Apple Silicon or Intel, Windows 10 64-bit+, Linux and Googlebook; CLI and IDE extensions.
- One Antigravity harness with plan-dependent Gemini, Claude and GPT-OSS models.
- Free, Google AI Pro/Ultra subscriptions, AI credits and Gemini Enterprise licenses are documented; API keys work in…
- CLI headless mode, a Python SDK and a preview managed agent API exist; webhooks and a hosted MCP server were not found.
- Teamwork captures requirements in a request artifact and maintains milestone plans and progress, with prompt…
- Automations (formerly Scheduled Tasks) and /schedule run prompts once or on a cron schedule; where they run and…
- Local execution, browser Remote Control of user hosts, and gradually rolling out Google-managed Remote Agents on…
- Local conversations can use project checkouts or Git worktrees, and the terminal sandbox uses native OS isolation.
- Headless print mode defaults to unlimited time with an optional --print-timeout.
- Per-model quota usage, headless token usage and credit balance alerts are documented; per-task cost is not.
- Free plan $0/month; Business Standard/Plus from $30 USD per seat per month for eligible Gemini Enterprise customers,…
- App v2.29.0 dated November 3 and CLI v1.5.1 dated November 5 are listed.
- Individual plan is $0/month with basic weekly rate limits and unlimited Tab completions; Pro and Ultra add limits…
- Pricing units are per seat per month or consumption; included limits are qualitative; subscription options require…
- The main Google Antigravity service and downloadable application use Google-authored service/software terms, with…
- Desktop threads can be imported into the CLI; 2.0 v2.19.1 can export rendered Markdown in artifacts as PDF; SDK…
- Desktop onboarding works with plain local folders or Git checkouts; hosted Git setup is not documented; a preview…
- MCP client, skills, plugins and hooks are supported; AGENTS.md and GEMINI.md are read, CLAUDE.md import is not…
- Antigravity 2.0 runs multiple local agents across projects and workspaces in parallel, with worktree mode and…
- Custom subagents and Teamwork roles support messaging, result return, live progress and follow-up; cross-harness…
- Steering, approval cards, resume, notifications and subagent stop are documented; pause semantics and error…
- Remote Control lists multiple machines and sessions with push attention alerts; 2.0 shows subagents in a panel and…
- Quick tasks, plan mode, Teamwork for large work and natural-language custom agents are documented; run-time role…
- Teamwork breaks a brief into milestones and parallel tracks; user-visible epics, blockers and readiness are not…
- Workspace code search, URL reading tools, MCP servers and image/Office inputs are documented; indexing and citation…
- Headless status values, approval cards, enterprise cloud telemetry logging, Remote Control daemon log locations and…
- Diffs, plans, walkthroughs, diagrams, screenshots and browser recordings are produced as artifacts.
- Agents run tests, hooks can run linters and block stops, and Teamwork adds reviewer gates; configured CI gates are…
- Enterprise SSO, Workforce Identity Federation and BYOID are documented; licenses are assigned by the organization's…
- Command, MCP, URL and file approvals with allow/ask/deny rules and presets exist; approver identity and revocation…
- Inline comments on plans and diffs, a Review Changes pane, walkthrough summaries and keyboard diff navigation are…
- Work starts from prompts in a local project conversation; external tracker intake was not found.
- Rules, AGENTS.md, skills and a /learn command provide shared instructions; memory behavior is not documented.
- Parent agents invoke subagents and Teamwork routes work, assigning each file to one worker; human assignment and…
- Sessions can be resumed, the Remote Control web UI tries to reconnect after network drops, and the headless Remote…
- Remote host setup via a headless daemon is documented; dependency and cache handling are not.
- Conversation histories persist across sessions and Teamwork records logs; export and integrity guarantees are not…
- Session history and cross-device access to the same session exist for the same account; colleague handoff is not…
- Worktrees, a Git-native VCS panel and local commits with generated messages are documented; pull-request features…
- /rewind rolls back conversation history; code revert semantics and delivery links are not documented.
- Individual use lets Google use Interactions to improve products with a settings opt-out; enterprise data is…
Sources
Open the Google Antigravity page
Claude Coderesearch correctionfreshness recheck51 findings · 63 sources
Findings today
- Claude Code is an MCP client (http, sse, stdio, websocket) with OAuth and scoped config, supports…
- Persistent instructions and memory are documented: CLAUDE.md at managed/user/project/local scope, AGENTS.md reading,…
- Outputs include diffs (/diff, Desktop diff view), live Artifact pages published to claude.ai, Desktop app preview,…
- Prompts and outputs go to Anthropic (or the chosen provider) over TLS 1.2+; consumer plans train on data only when…
- Claude Code documentation points to the Anthropic Trust Center for compliance artifacts.
- TaskCreate, TaskGet, TaskList and TaskUpdate manage session checklists.
- Live multi-human collaboration is limited.
- Claude Code has CLI, Desktop, IDE, browser and mobile clients.
- Compute-aware scheduling is limited: Remote Control and self-hosted runners cap concurrency, Bash commands have…
- The installed Claude Code CLI can run its full agent loop, tools and context management on user-controlled…
- Public seat prices, plan multiples and separate API rates are stated.
- Claude Code embeds its own harness and runs Claude models only.
- Pro, Max, Ultra, Team and Enterprise include Claude Code.
- CLI onboarding is one install command plus browser login, and works in any directory.
- Programmable access is strong on the CLI: claude -p with JSON output, Python/TypeScript Agent SDK, claude mcp serve…
- The mobile app is a client for cloud or Remote Control sessions, and IDE extensions wrap or bundle the CLI engine.
- Parallel work is supported locally (multiple terminals, worktrees, subagents, agent view background sessions, Remote…
- Agent teams (experimental) document lead assignment and file-locked task claiming; outside teams, assignment is the…
- Delegation is inspectable through subagents (own model, effort, tools), agent teams, cross-session messaging and…
- Cloud scheduled tasks, formerly routines, support schedule/API/GitHub triggers independently of the laptop.
- In the CLI a human can interrupt, stop, queue steering messages and resume sessions; background sessions show Needs…
- Agent view (research preview) is a terminal fleet screen that groups background sessions under Needs input, Ready…
- Quick tasks are one prompt away; larger efforts use subagents, agent teams, projects (cloud beta) or dynamic…
- Claude Code is proprietary: the public repository LICENSE.md states "© Anthropic PBC.
- Pro $20/month or $200/year; Max $100/month; Ultra $200/month.
- Fixed subscription examples: one Pro $20/month or $200/year; two Team Standard seats $50/month or $480/year; five…
- Current changelog lists v2.1.319.
- Free excludes Claude Code; all paid plans include it.
- No native issue tracker was found in the documentation, changelog or repository checked; work arrives by prompt,…
- Plan mode proposes a plan before edits and plans can be saved into the repository; /goal tracks a completion condition.
- Ordinary TaskUpdate changes task dependencies.
- Code understanding relies on agentic search (grep, glob, file reads) plus optional LSP code-intelligence plugins and…
- Claude Code has persistent native task checklists with optional cross-session lists and experimental team…
- Living product specifications were not found as a native feature; instructions files and plans in the repository are…
- Claude Code runs locally by default; cloud sessions run on Anthropic VMs (4 vCPU, 16 GB, 30 GB) with no separate…
- Isolation options: git worktrees per session, an OS-level sandbox for shell commands (macOS/Linux/WSL2, off by…
- Sessions persist as local transcripts and resume; cloud sessions restore history after VM reclamation but lose…
- Cloud environments offer setup scripts, a filesystem-snapshot cache, network access levels and pre-installed…
- Limits include print-mode --max-turns and --max-budget-usd (client-side estimate), subagent maxTurns, tool timeouts,…
- Operational statistics are available via /usage, /insights, OpenTelemetry metrics, logs and traces (beta), and the…
- Cost views: /usage session cost (client-side estimate at list price unless an administrator's managed modelPricing…
- Verification is policy you configure: hooks run commands and can block actions or task completion (exit code 2),…
- The CLI supports git workflows: it can create commits and pull requests on request, run in isolated worktrees, link…
- Merge and deployment authority is delegated to git/CLI permissions: Bash allow, ask or deny rules can gate push and…
- Completion evidence is session links to PRs and OTel commit/PR counts; rollback is via /rewind checkpoints on local…
- High-volume shipping features are limited: agent view summaries, /batch splitting a change into worktree subagents,…
- Roles and access are managed at the Claude account level: Owners and admins configure cloud environments and…
- Approvals are per-user permission prompts governed by permission modes plus allow/ask/deny rules, hooks and…
- History comes from local session transcripts, checkpoints, OpenTelemetry events and, on Enterprise, audit logs.
- Review is available through /diff, Desktop diff view with inline comments, local /code-review, and managed GitHub PR…
- Takeover is mostly same-user: teleport and Remote Control move a session between one person own devices; colleagues…
Sources
Open the Claude Code page
Conductorfreshness recheckresearch correction44 findings · 43 sources
Findings today
- Conductor documents a macOS app and an iPhone app requiring iOS 26.0+.
- Subscriptions and API keys are accepted locally and in Cloud; Sign in with ChatGPT was added for Codex; Conductor…
- Git-backed setup with credential checks; GitHub required; Cloud onboarding is guided and BYO cloud machines are…
- Beta HTTP API (/v0, OpenAPI), hosted Streamable HTTP MCP server and routine webhooks; Conductor CLI exists in cloud…
- API/MCP (beta, cloud workspaces) can create sessions, send prompts, cancel the current turn and read transcripts; an…
- Cloud routines run saved prompts on schedules (including weekdays), GitHub actions with filters, or webhooks;…
- Cloud scope.
- Managed Cloud (Vercel sandbox, us-east-1) keeps agents running after the Mac app closes until sandbox sleep or…
- Experimental cloud cost estimates (Codex, 0.77.0) exist; per-task cost reporting is otherwise not documented;…
- Shared Cloud workspaces support live chats, presence and typing indicators; private workspaces exist; Pro…
- Cloud organizations have admins and members.
- The local app keeps chats and files on the Mac (~/Library/Application Support/com.conductor.app and ~/conductor),…
- The pricing URL redirects to the homepage without a price table.
- One-, two- and five-person scenarios cannot be priced from the checked public pages because current subscription…
- Conductor has withdrawn its wind-down plan.
- Pricing is partially specified: model billing stays with the provider and Cloud FAQ states no additional compute fee…
- Plan Mode for Claude and Codex and workspace todos organize longer work.
- Local tool approvals exist, with settings for defaults, plan mode and managed enforcement.
- Four agent harnesses (Claude Code, Codex, Cursor, OpenCode) with several model-provider routes; Claude Code, Codex…
- MCP client use through each harness with user/project scopes, OAuth and approval prompts; instruction files and…
- GitHub and Linear issues, branches and PRs can start workspaces; prompt links prefill the composer.
- Repository settings, instruction files and skills carry shared guidance; persistent memory is not documented.
- Attachments of notes, specs, screenshots and logs are supported, and file-content search exists (0.87.0); indexing,…
- Workspace todos exist; an independent issue backlog is unverified.
- Agents can receive workspace todos; shared native issue operations remain unverified.
- Multiple workspaces, or several chats in one workspace, run in parallel; Home and the sidebar show workspace status,…
- Cloud workspaces can be reassigned to teammates; agent assignment is not documented.
- Home filters and sidebar sections organize workspaces; a matrix-style status indicator was introduced as…
- Per-session harness, model, Fast Mode and reasoning-level controls exist; workflow presets, coordinator and…
- Local workspaces use a git worktree and branch each without an OS sandbox; Cloud uses isolated Linux microVM sandboxes.
- Cloud preserves files and chat history across sleep but not running processes; interrupted turns must be resent.
- Cloud Computer builds seed new environments; setup scripts run per workspace; port forwarding is off by default for…
- Cloud sandboxes enforce sleep and lifetime limits; Codex chats retry on OpenAI rate limits (0.84.0, no surface named).
- API session status distinguishes idle, working and errored; failed Cloud startup exposes setup logs.
- Teammates can continue and reassign a Cloud workspace.
- Cloud workspaces show diffs; plans, screenshots and attachments in .context sync from cloud workspaces to the Mac…
- Checks aggregate CI; unresolved todos block merge; the checks page says Conductor "may block or discourage" merge…
- Workspace branch-to-PR workflow is documented, with GitHub stacks (0.80.0; local needs the gh stack extension) and…
- One-click PR approval and merge from Conductor; merge blocked if the PR changed on GitHub.
- Local checkpoints support destructive turn reversion; the Checks tab may include pull request metadata and deployments.
- Conductor terms explicitly define the client-side and hosted Platform as proprietary software.
- Local workspaces and chats stay on the Mac, model requests go to the configured provider, and analytics go to PostHog.
- Conductor reports SOC 2 Type II; audit report and scope not inspected.
- Diff viewer, line comments, GitHub review threads and suggested PR actions guide review.
Sources
Open the Conductor page
Cursorresearch correctionfreshness recheck48 findings · 53 sources
Findings today
- Cursor is reachable from a desktop app (Mac, Windows, Linux), a CLI, a web client at cursor.com/agents, a native…
- Cursor runs its own coding-agent loop with selectable models.
- A Cursor account uses the new X Subscription or a retained legacy Cursor plan.
- Cloud Agents need a paid plan and connected source control (or a no-repo start); setup is agent-led or…
- Cursor exposes a CLI with headless mode, a public-beta Cloud Agents REST API, a TypeScript SDK, and signed status…
- Work is routed by Slack keyword rules and by Jira repository-name, service-name and keyword matching; Team Pool…
- Privacy Mode prevents training.
- Fixed new-plan examples only: one Premium user $30/month; two Premium Teams seats $60; five $150.
- Cursor announced its completed acquisition by SpaceX on August 14, 2026 and described continued product development.
- Work can start from prompts in the app, CLI or API, and from Slack, GitHub, Bitbucket, Linear and Jira mentions or…
- Agents interact with external Linear and Jira issues (assigned or mentioned) and report progress there; a native…
- Local agents use worktrees and an OS sandbox; cloud agents run in per-agent Firecracker microVMs with configurable…
- Cloud agents accept follow-up runs and CLI sessions can be resumed.
- Cloud runs produce diffs, PRs and artifacts (screenshots, videos, log references) with remote desktop takeover and…
- Agents open pull requests (ready for review unless a draft is requested).
- Cursor is an MCP client with OAuth and admin allowlists, supports plugins and marketplaces, and reads skills,…
- Subagents (built-in and custom) and Project coordinators delegate work with isolated context and return final…
- Automations run on schedules, source-control events, Slack, Linear, Sentry, PagerDuty and webhooks; Projects can…
- Cursor is a vendor-hosted service; only tool execution can be self-hosted, and cancellation reverts the account to…
- Cursor's pricing footer displays "SOC 2 | ISO27001 | ISO42001 | AIUC-1 Certified" (vendor statement, pricing page…
- New-plan seat rates and annual prices are public; usage pools are weekly and third-party models prepaid.
- Cloud setup builds reproducible environments from agent-led setup, snapshots or Dockerfiles with install/start…
- Agents are visible to members of the team they were started under and are read-only for viewers; follow-ups are…
- Enterprise audit logs record admin and configuration events with CSV, API and SIEM export; conversations persist…
- Users can steer, queue, cancel, resume and follow agents from several clients; notifications reach mobile and Slack.
- Cursor supports one-off tasks in Agent mode, longer coordinator-led Projects, Custom Modes with skills, and subagent…
- Teams have Member, Admin and Unpaid Admin roles with SAML SSO; Enterprise adds SCIM, organization groups and MDM…
- Run modes, sandbox rules, MCP approvals and hooks control agent actions locally; cloud agents run commands without…
- Many agents can run in parallel across local worktrees, remote SSH machines, self-managed machines and the cloud; no…
- The Agents Window and the mobile app list agents; Projects show the agents their coordinator started.
- The Terms grant a limited right to access and use the Service and reserve all rights not granted; the public…
- New subscriptions cost $8 Plus, $30 Premium, $100 Super or $200 Ultra monthly; Teams seats $30/$100/$200.
- Free has limited core use.
- Plan Mode produces editable markdown plans before building; acceptance criteria, revision history and plan-to-work…
- Projects let a coordinator agent plan larger bodies of work and delegate to cloud agents; task dependencies,…
- Rules, AGENTS.md, skills, team rules, automation memories and Project shared-context files carry instructions and…
- Agents search code with a local grep index, web search and MCP, and accept image inputs; embeddings-based indexing,…
- Work can run on the local machine, on remote SSH hosts, on Cursor-managed VMs, or on self-managed workers; managed…
- Spend limits bound cloud usage; time limits and timeouts are not stated.
- Logs, run states, per-run token usage and admin analytics exist; traces and approval-state reporting are not documented.
- Token usage is available per agent and per run through the API and SDK; billed dollar cost per agent is available…
- Team Pools queue requests for workers and a controller scales machines; other compute-aware scheduling and test…
- Independent AI review (Agent Review, Bugbot) and policy files exist, and Bugbot can fail a status check; lint, type…
- Cloud agents push branches, make signed commits, open draft PRs and respond to PR comments; local work uses worktrees.
- PR, deployment and check status appear in mobile, and Rollouts monitors deploys for regressions and notifies the…
- High-volume batch review views were not found; mobile and PR routing give partial support.
- Agent Review, Bugbot, a diffs view and mobile PR review support reviewing agent output; shipping guidance and risk…
- Teammates can continue shared cloud runs when an admin enables team follow-ups, take over the agent desktop, or fork…
Sources
Open the Cursor page
JetBrains Airresearch correctionfreshness recheck41 findings · 58 sources
Findings today
- Air Teams automations can be started by an incoming webhook and can send outgoing webhook notifications; the Central…
- Task context can reference files, folders, symbols, Git branches and commits, local changes, terminals and uploaded…
- A task starts from text typed in the task input.
- Run environment is selectable: Local Workspace (no isolation), Git Worktree, Docker (local) and Cloud; worktree and…
- Air Teams uses organization membership and project admin/member roles.
- Cloud tasks commit to a task branch and push after every agent turn; a person then opens the pull request from that…
- Before a commit, the Task Changes tab and the Changes tool revert changes file by file or all at once.
- The EAP agreement says JetBrains does not see or access user Data in the product by default without consent and that…
- Air is reachable through the Air app on macOS, Linux and Windows, Air Teams in the browser, and Air in JetBrains…
- Air in IDEs runs local coding agents through ACP and terminal surfaces, explicitly including Claude Code, Codex and…
- First local task needs Git, the Air app and one connected provider; cloud tasks need an organization set up in…
- Air supports MCP in the Air app (global scope available in all projects on the machine, local scope in .air/mcp.json…
- Tasks are created from a prompt in the Air app or browser; automations can start from GitHub, Jira or webhook events.
- Plan mode lets the agent write a plan as a Markdown file in .air/plans before implementation and the user approves…
- Air picks up CLAUDE.md and the .claude folder for Claude Agent, reads AGENTS.md for agent workflows, and loads…
- The Air app runs several tasks at once across projects, Air Teams runs each cloud task in its own environment, and…
- A review pass can run in a separate agent session with a different agent and model and its accepted comments are…
- Air Teams automations run in the cloud on a schedule, at an interval, or from GitHub, Jira or webhook events without…
- Automations can be paused and resumed and a run in progress can be stopped; suspended cloud environments and…
- Air app tasks group by date, status or target and flag Input required.
- Air in IDEs runs agent sessions on the user's machine by default, with Local and Temporary Worktree modes, and Air…
- Cloud tasks keep running after the browser tab closes.
- Cloud environments use a startup script, snapshots and shared configurations; worktree and Docker tasks use…
- Documented limits are the monthly credit limit, a startup script timeout (default 600 s, 1-3600 s) and inactivity…
- Cloud tasks expose a Problem status with the error at the top of the task page, downloadable logs (Collect Logs and…
- In the documented current release, Air Teams usage tracking covers JetBrains AI only, and the Air app credit counter…
- Air Teams projects share environments, connectors and automation results.
- The Air app offers Plan, Ask, Edit and Full Access permission modes; Ask requires approval before changes or…
- Central Console keeps an immutable audit log of organization administration changes (not AI activity) with CSV…
- The Air app shows a merged Task Changes diff with unified or side-by-side views, line comments and Add to Task…
- Diffs are reviewed in the Task Changes tab and their commits remain on the task branch; plans are Markdown files in…
- In the Air app, agentic review runs in a separate agent session; the browser has no built-in review pass.
- The Comments tool can show comments for the current task or across tasks, diff stats summarize size, and Tasks tool…
- Air is proprietary: the EAP User Agreement (v1.0) section 3.3 prohibits reverse engineering, decompiling, trying to…
- The Air app can run local tasks without Air Teams or an organization on the user's own provider account.
- Console Business is $20 per seat per month (monthly or annual billing); 1 AI credit equals 1 USD of provider-side…
- Air is in early access: the Air app is under an EAP user agreement, Air in IDEs is EAP, Air Teams is available to…
- JetBrains states SOC 2 Type II covers the Central Console and its governance layer and that all plans are GDPR…
- Air app EAP terms allow free app use; AI usage is separate.
- Units are AI credits (1 credit per 1 USD of provider-side LLM cost) and Business plan AI seats at $20 per seat per…
- Cloud tasks push changes to a task branch.
Sources
Open the JetBrains Air page
Multicafreshness recheckresearch correction49 findings · 65 sources
Findings today
- Web app, Desktop app (macOS, Windows, Linux) and a CLI exist; the iOS client is self-built from source, with no App…
- The vendor website, the providers table and the install table each list 29 supported coding tools, including…
- Multica does not provide model access in the sources read: each coding tool uses the account, subscription or API…
- Documented onboarding is: sign in, connect a computer with Desktop or multica setup, create an agent, assign the…
- The CLI exposes broad issue, agent, project, autopilot and runtime operations over the shared server API,…
- Workspace owners/admins keep an MCP server library that is assigned per agent and passed to most supported tools;…
- Desktop uses Electron; mobile docs specify React Native.
- Multiple runs execute concurrently on a daemon (default 20 per daemon and 6 per agent) and work state is visible in…
- Issues can be assigned to members, agents or squads; runs queue for a runtime and are claimed by it; consecutive…
- Squads provide product-owned delegation: an agent leader receives the issue, @-mentions member agents, records an…
- Autopilots run on 5-field cron schedules or webhook triggers from the Multica server, and issue wakeup rules can…
- Run history exposes stop and retry; running Claude Code, Codex and Grok tasks can be steered with new instructions;…
- Issues can be browsed in list, board, table, Gantt and swimlane views with status, assignee and project filters,…
- No mandatory workflow preset is required; a quick chat or issue works alone, work can grow into sub-issues, and an…
- Custom Multica License includes extra conditions.
- Multica can be self-hosted (Docker Compose, Kubernetes) with Desktop and CLI pointed at it; self-hosted use needs no…
- Workspace content (issues, comments, run records, attachments) is stored by Multica; code and tool credentials stay…
- No public price list was found: /pricing renders a not-found page and the site offers Get started, Book a demo and…
- Internal self-hosted software license cost is $0 for one, two or five users.
- Current release is v0.8.1 dated November 6.
- Self-hosting is free for internal use (infrastructure and tool accounts are the user's), the website offers a Start…
- A commercial license for external hosted/embedded use is quote-only.
- Multica issues are the native record: issues are created in the app, CLI, chat bots or by prompting an agent.
- Issues hold the task brief and requirements (description text, comments, attachments); no separate structured plan…
- Issues have parent-child relationships.
- Agent instructions, workspace context, project descriptions and reusable skills provide shared context; sessions…
- Multica passes attachments and chat media to agents but no product-level code search, indexing, architecture…
- Native issues include creation, editing, comments, priority, labels, assignment, search, filters, projects, five…
- Agents use CLI against shared issue records.
- Connected computers execute local-tool runs while Cloud or self-hosted Multica coordinates them.
- GitHub repository runs get their own worktree from a cached bare clone; local git directories can use per-run…
- Transient failures auto-retry (default 2 attempts, 3 for tool network errors) and a daemon restart reclaims…
- Repository checkouts use cached bare clones and per-task worktrees; the user supplies and maintains tool installs…
- The daemon enforces a configurable absolute run time limit (default none), an idle watchdog and a tool watchdog…
- Execution log exposes run state, timing and streaming transcript with classified failure reasons; a workspace…
- Estimated per-run and per-issue cost appears in the execution log with a cumulative curve, and the CLI reports issue…
- A daemon-wide cap (20) and per-agent cap (6) queue excess runs, and a directory lock serializes runs on the same…
- Issues retain comments, attachments with versions, a deliverables sidebar and in-place previews (HTML, Markdown,…
- Multica shows CI status and mergeability on the PR card and can move an issue when linked PRs merge, but no…
- Agents run Git and the GitHub CLI themselves; Multica adds per-task worktrees, a task Git identity, optional…
- Merging is done by humans or by agents on request through the user's own Git tooling; the GitHub App is read-only.
- Linked pull requests show state and merge status on the issue and merged PRs can complete it; no build, deployment,…
- Review happens one issue at a time through the inbox, in_review status and PR cards; batch queues, grouping, risk…
- Workspace members share issues and comments with attribution; members can each comment, mention and steer agents…
- Workspaces have owner, admin and member roles, 7-day invitations and immediate revocation on removal; agent Access…
- Runs are unsandboxed and tool approval prompts are answered automatically by default; agent Access controls who can…
- The issue timeline records attributed activity and comments with actor type and id, comment revisions and resolve…
- Review is issue- and PR-centric: in_review status, a pull request card with state, size, CI and mergeability,…
- Issues, comments, transcripts and attachments live on the server, so another member can follow the same issue and…
Sources
Open the Multica page
Orcafreshness recheckresearch correction51 findings · 44 sources
Findings today
- Desktop on macOS, Windows and Linux, plus beta iOS/Android companion.
- Orca launches and tracks existing CLI agents in terminals rather than embedding a harness: many CLI launchers are…
- Orca reuses the installed coding-agent native subscription logins, including Claude Code and Codex.
- Documentation describes install, first launch (home-directory access, optional import of Claude, Codex and Ghostty…
- Orca exposes a broad orca CLI (worktrees, terminals, files, browser, orchestration, automations, Linear, accounts,…
- Orca registers MCP servers for MCP-capable agent CLIs, installs agent skills, imports skill folders from Claude,…
- Source identifies an Electron and React desktop app and an Expo and React Native mobile app; docs mention a…
- Work can start locally from a name or prompt, or from GitHub, GitLab, Linear or Jira items; the external tracker…
- Orca keeps lightweight task state: orchestration tasks carry a spec and status, worktrees carry a free-text comment…
- Workspaces can be nested under a parent workspace and orchestration tasks have dependencies and statuses including…
- Orca relies on the agents own instruction files (CLAUDE.md, AGENTS.md) and repo hooks, offers installable and…
- Multimodal inputs are documented: files and images dragged into agent prompts, Design Mode sends element HTML, CSS…
- Orca documents native Run/Task/Dispatch records independent of external trackers, but the required Experimental…
- Agents can read, create and update Linear issues through orca linear and set worktree comments and status, report…
- Native orchestration tasks retain a spec and link to execution attempts and decision gates.
- Multiple agent sessions run concurrently, each in its own worktree and terminal, with status visible in the sidebar,…
- Orchestration dispatches a task to a specific worker terminal and treats the active dispatch as completion…
- Orca documents its own orchestration skill/CLI and supervised worker delegation, but the required Experimental panel…
- orca automations run a prompt on hourly, daily, weekday, weekly, cron or RRULE schedules against a repo or worktree,…
- Users can stop, steer, queue messages, answer pending questions and approvals (in terminal or Chat UI), and are…
- Fleet views documented: sidebar worktree cards with N-children chips, an experimental kanban dashboard (Needs You,…
- Creating a worktree and launching an agent is a create-and-launch sequence with no mandatory workflow selection (the…
- Local desktop, user-owned SSH hosts and persistent Remote Orca Servers; experimental Cloud VM recipes use the user…
- Each task is a real git worktree with its own branch and directory.
- A background daemon keeps local agent processes running across app quit, update relaunch and Orca app crashes.
- New worktrees can run setup hooks, share gitignored directories such as node_modules by symlink, and copy .env-style…
- No enforced time, turn, token or spend limit was found.
- Agent state is shown from hooks and OSC titles with a distinct waiting-on-you state, a feed and a dashboard; Stats…
- Orca shows provider usage windows, resets and credit balances, and Stats with estimated cost from a local price…
- A Resource Manager shows CPU, memory and session counts with daemon controls, and idle agents can hibernate to free…
- Orca is built around one developer machine or server; several paired clients can see the same server-owned workspaces.
- The self-managed Orca runtime uses revocable paired-device grants without named organization/project user roles or…
- Approval handling is largely delegated to the agent CLIs: by default Orca pre-fills each supported CLI's…
- Orca keeps local per-line attribution of human versus AI edits, a feed of agent events, searchable on-disk session…
- Review is built into each worktree: a combined diff, image and HTML preview, three-way conflict resolution, line and…
- A second client paired to a Remote Orca Server reaches the same server-owned workspaces, tabs and agent sessions.
- Diffs, HTML and Markdown previews, per-worktree browser pages, screenshots through the CLI and shared HTML or…
- Repo pre-commit hooks run on commit with an inline failure and a Fix with AI repair prompt, failing GitHub checks…
- Branches, worktrees, staging by hunk, commits, push, PR or MR creation (GitHub, GitLab, Bitbucket Cloud) with draft…
- Orca can merge or enable auto-merge on GitHub pull requests using the repository allowed merge methods and can merge…
- Worktrees link to pull requests and merge requests with live status, and CI check results appear on the worktree.
- A done queue exists in the experimental dashboard and Agents feed with response previews, PR status filters and…
- The public repository carries a root MIT license that covers the desktop app, mobile app, CLI and the cloud/ Relay…
- Orca can be fully self-hosted for desktop, SSH and Remote Orca Server use with no Orca account; Relay pairing,…
- Vendor policy describes local code/provider credentials and anonymous opt-out telemetry.
- The Orca application is advertised as free and open source with no price list; the Enterprise page offers rollout…
- No priced scenario was built.
- Latest GitHub release is v1.4.253, published November 7, 2026, following v1.4.252 the same day and v1.4.251 on…
- The Enterprise page shows an "AICPA SOC 2 Readiness" label and the question "Is Orca SOC 2 compliant?", The captured…
- The desktop app is advertised as free and open source, so BYO-subscription and self-hosted scenarios carry no Orca…
- The site advertises the desktop app as free; no desktop seat, usage-unit or overage rate is established by the cited…
Sources
Open the Orca page
Paseofreshness recheckresearch correction47 findings · 34 sources
Findings today
- Paseo offers desktop, native phone, browser and CLI clients for its daemon; the shared browser app includes compact…
- Current stable and homepage document native Claude Code, Codex, OpenCode, Pi, OMP, Antigravity and Muse plus a…
- Paseo reuses each provider CLI's own login for Claude plan, ChatGPT plan and API-key use, and adds no charge for…
- Paseo works in any directory on a daemon host, with or without git; worktree features and PR workspaces need a git…
- Paseo exposes a CLI, TypeScript SDK, daemon MCP tools and a Hub HTTP API with scoped keys and OpenAPI; daemon access…
- Paseo documents a plugin system and passes through the provider CLI's own MCP servers and skills; it does not…
- Paseo desktop is an Electron app; the mobile and web clients share an Expo / React Native app package.
- Paseo is workspace- and session-based and is not documented as having a native task-tracker intake; agents start…
- Paseo documents plan mode as a session mode and persists agent timelines; a durable task plan or acceptance-criteria…
- Paseo models parentage between agents (subagents, detach) and workspaces; epics, blockers and readiness states are…
- Paseo relies on each provider CLI's instructions, skills and memory; it adds installable orchestration skills and…
- Paseo documents file browsing, a built-in browser and screenshot tools; code search, indexing and retrieval are not…
- The inspected release exposes agent workspaces and provider task displays, but no native issue tracker.
- Paseo runs multiple agents and workspaces concurrently, each in its own worktree if wanted; documented recipes cover…
- Paseo documents agent parentage and per-agent addressing; human assignment, agent claims and queue routing are not…
- Agents can launch, prompt, inspect and cancel other agents of any provider; workers are retained as inspectable…
- Paseo schedules start new agents on a cron cadence (with run and expiry limits) and heartbeats re-prompt an existing…
- Users and parent agents can cancel a turn, stop or kill an agent, send follow-ups, change mode, and answer…
- Paseo has a project/workspace sidebar and Subagents track but a fleet board with dependencies and workload views was…
- Paseo offers profiles, skills and natural-language orchestration; richer workflow flexibility questions are…
- Paseo runs agents on operator-owned local or server daemon machines.
- Paseo isolates file changes with optional git worktrees per workspace; same-workspace agents share files.
- Paseo persists agent records and timelines and resumes closed agents under the same ID.
- Paseo worktree setup hooks, scripts and supervised services with port allocation address environment setup; caching,…
- Paseo documents run counts and expiry on schedules and max_runtime / idle_timeout on Hub steps; token, spend and…
- Paseo provides agent timelines (paseo logs), daemon and desktop logs, a Usage view of subscription quota windows and…
- Paseo shows each subscription account's quota windows and context-window usage; per-task or per-epic dollar cost,…
- Paseo documents no compute-aware scheduling; Hub dispatch is not queued and agents are launched on demand.
- In self-hosted Paseo Hub, distinct permitted Slack users can send instructions to one daemon agent by enabling Same…
- Paseo Hub has owner, admin and member organization roles with enforced management permissions.
- Paseo routes provider permission prompts to the app, CLI and MCP; Hub runs use the provider permission modes set on…
- Paseo persists agent timelines and Hub records every accepted event and run; export, retention and integrity…
- Paseo provides a Changes (diff) tab, Files, PR panels with check summaries and split panes; a pending-review queue…
- Another user with pairing credentials can connect to the same daemon and, with workspace.read, see its workspaces…
- Paseo shows workspace diffs and files, previews apps in built-in browser tabs and service proxy URLs, and browser…
- Paseo lets users define named workspace scripts (tests, lint) and run them from app, CLI or agents; required checks,…
- Paseo creates worktrees and branches, drafts commit messages and PR text, opens PR workspaces and detects linked…
- Merge and deployment authority are not documented by Paseo; Hub steps limit GitHub authority via declared…
- Paseo documents PR links for workspaces; commit, build and deployment links, completion evidence and code rollback…
- Paseo v0.15.3 root code is Apache-2.0 with third-party licence exceptions; its desktop manifest also declares…
- Paseo runs entirely on user machines, with optional relay and Hub; self-hosting Hub is documented.
- Stable v0.15.3 states no telemetry, tracking or required Paseo account for the local app.
- The Paseo app, daemon, CLI and SDK are free; hosted Hub is Free ($0, 50 agent runs/month, 1 seat) or Pro ($15 per…
- Fixed hosted Hub examples: one Free seat $0 for 50 runs/month; two Pro seats $30/month; five $75/month.
- Current stable is v0.15.3 on November 6, with separate Beta channel and gradual desktop rollout.
- The Paseo app, daemon, CLI and SDK are free and open source with no usage cap documented; hosted Hub offers a…
- Hub pricing lists the unit (seat), included runs and the seat definition; Paseo documents no model or compute charge…
Sources
Open the Paseo page
Podiumproduct changeresearch correctionfreshness recheck48 findings · 101 sources
Findings today
- Podium ships desktop, CLI and browser clients, including phone-browser control.
- Docs describe a first-run flow (install, podium, register a Git repository, confirm a detected logged-in harness)…
- v0.3.0 routes Podium issue instructions and relay environment through terminal and engine session launch.
- Podium exposes native issue creation with an optional start flag, separate dispatch, editing, comments, labels,…
- Chat and native CLI views share a session.
- Closing a browser tab does not stop the agent process on a running Podium host.
- Podium exposes retryable errors, optional auto-continue with backoff, bounded waits and provider quota.
- CLI and MCP address the same canonical issue records for reads, creation, edits, claims, state, dependencies and…
- Podium defines member/admin instance roles and admin-only membership management.
- Local desktop and self-hosted installs run on user-controlled machines with a documented state directory.
- Podium launches the native Claude Code, Codex, Grok, OpenCode, Cursor and Pi CLIs in persistent terminals on…
- Podium supports native coding-agent logins and managed provider credentials.
- The podium CLI covers issue create, show, update, claim, dependencies, artifacts, session send/status/read/stop and…
- Podium gives agents its own issue tools over CLI and MCP, and agent profiles import and export Claude Code agent files.
- Desktop releases use a Tauri 2 shell that serves the Podium web UI; mobile source uses Expo and React Native.
- Podium issue records are native and authoritative with stable references, CLI creation and editing, and agent-driven…
- Issue brief, acceptance criteria, state paragraph and todos persist on the durable issue record.
- Podium supports persistent parent and child issues, phases and typed dependency links, including reparenting an…
- Not established.
- The spec CLI and service (living specs under pspec with stable SP IDs) are in v0.3.0.
- Docs and release notes describe many sessions running as independent terminals, one worktree per issue, and several…
- Issues carry an assignee, claim sets assignee and in_progress, a coordinator session can be set per issue, sessions…
- Podium orchestrates delegation itself: podium agent spawn starts a full child session on an issue with a chosen…
- The server owns a scheduler for cron and one-time automations that spawn agent sessions, skipping overlaps and not…
- Release 0.3.0 ships a flight deck showing a whole mission in one view: phases, every task and its state, who works…
- A quick task can be an interactive session or an issue; an issue can grow into a mission by gaining children while…
- The published Podium execution choices are local user machines and a user-owned VPS or paired machine.
- Podium separates issue files with Git branches and worktrees, but its released daemon launches native host processes…
- Only machine setup is documented: the installer installs missing runtime tools and pairs a host, and the state…
- Podium offers local logs and crash export, structured session status, issue stats, a usage and analytics sheet with…
- Since 0.2.0 Podium shows what each task costs, on the task header and the usage sheet, with per-session rows and…
- Podium reports host memory, load and disk, auto-hibernates idle sessions under pressure, and provides named advisory…
- The issue UI displays owner and visibility read-only and deliberately omits sharing controls.
- Agent-session management requests go through an operator approval broker with approve and deny; the privacy docs say…
- Issue actions, approval steps, comments and mail are recorded server-side.
- Review is a documented loop: issue stage review, diff and checks, artifacts, offers with merge or send-back buttons,…
- The durable issue, session transcript, artifacts and mail persist for any later agent, and a live session can be…
- Artifact commands store immutable server-side copies of issue-linked files, including images, video, HTML and…
- In the experimental Shipping path a configured validation profile runs before landing (default argv bun run test…
- Issues own branches and worktrees, and issue actions rebase, open a pull request with gh (pr create --fill) and…
- The operator accepts work through offers and runs the merge action under a branch merge-lock; in the experimental…
- Shipping orders record approved base and head SHAs, tested integration SHA, landed ref and validation result in a…
- The shipping service records revision-specific approval, configured validation and delivery receipts.
- The public repository is licensed Apache 2.0, and the website presents Podium as an open-source agent development…
- Docs state repositories and native harness credentials stay on machines you control and server state lives in a…
- No Podium price is published: podium.do/pricing returns 404 and the docs list billing as not implemented.
- Podium is pre-1.0.
- Podium is open-source Apache-2.0 software, and the download page calls it free and open source.
Sources
Open the Podium page
Paradevfreshness recheckresearch correction47 findings · 49 sources
Findings today
- Paradev, formerly Superset, has macOS desktop, experimental Linux, CLI and iPhone/iPad clients.
- Paradev documents 22 built-in terminal agents, custom CLI agents and mixed-harness orchestration.
- Superset does not sell model access: all plans use the user's own agent subscriptions or API keys, and it does not…
- Onboarding detects git, gh and installed agents; projects come from a local folder or Git URL and must be git…
- Current CLI and OAuth HTTP MCP tools cover tasks, workspaces, agents, terminals, pages, automations and hosts.
- Current Paradev plugins supply MCP tools and skills with agent-specific support and a vendor credential proxy.
- Native tasks and connected Linear/GitHub issue lists support workspace intake.
- Task projects group tasks with state, lead and dates in the default-branch source (source-inspected at fc44d327:…
- Bundled Superset skills are installed into agents, and a session can be handed to another agent with its last 36,000…
- Design mode sends screenshots and DOM context to agents and the Slack agent searches the web with citations; code…
- Native tasks support create, edit, priority, assignee, labels (with label hierarchy), semantic search,…
- Agents can list, read, create, update and soft-delete tasks over MCP, CLI and SDK.
- Paradev documents parallel agents in separate local worktrees, with sidebar activity and workspace list/board views.
- Tasks have assignees and can be dispatched to a chosen agent; the Linear script reuses an existing workspace per issue.
- Paradev bundles the superset:orchestrate skill: a Claude Code or Codex coordinator launches mixed-agent workers via…
- Current automations combine schedules and provider-event triggers, up to 25 per automation, on selected hosts or cloud.
- Agents can pause for questions, tool approval and plan review; workspaces needing input are flagged and notified;…
- A workspace board groups work by agent and PR state; parent/child and dependency fleet views were not found.
- A quick task takes a prompt and an agent in one step; agents are configurable and skills are discoverable.
- Local and user-hosted remote execution plus managed Team/Ultra Cloud workspaces.
- Local workspaces are git worktrees on separate branches; the FAQ states they are not sandboxes and share processes,…
- Terminals run in a background daemon that survives app restarts and updates; unexpected agent-terminal deaths…
- Setup and teardown commands run per workspace on create and delete (install dependencies there).
- Usage shows local Claude Code and Codex token cost, quotas and per-model/workspace breakdowns; automation run…
- Usage estimates token costs from local session logs across supported agents, with workspace/model/session drilldown…
- A live resources view shows CPU, memory and processes per project and terminal; scheduling and verification controls…
- Granted organization members can open the same remote host workspaces and send input to their shared agent terminals.
- Remote hosts have explicit member grants with owner promotion and removal; Enterprise lists SAML SSO and SCIM.
- Agents can pause for tool approval in the chat pane.
- Sessions, run history and task activity persist; export, retention and integrity are not documented.
- Pull Requests and diff views offer split/unified/focus diffs, check status, review threads and iPhone diff review.
- Granted teammates can open authorized remote workspaces through the relay; separately, a session can be handed to…
- Paradev documents split/unified and binary diffs, versioned Pages with workspace lists and retrieval, browser…
- Check status is displayed on PRs; configured verification gates were not found.
- Built-in diff and Pull Requests views stage, commit, push, create PRs and merge (squash, merge commit or rebase and…
- PRs can be merged (squash, merge commit, rebase and merge) from Superset on the real PR through the user's GitHub…
- PR-based workspaces show pull request state in the sidebar, and an Open Preview button appears when the branch has a…
- The Pull Requests view lists PRs with check counts, and attention-based board columns (Needs attention, Needs…
- Current desktop-v2.0.5 root repository code is MIT-licensed.
- The MIT local core can import a repository, run agents and review diffs signed out and offline on the user machine,…
- Paradev discloses desktop analytics.
- Per-seat Pro $20 monthly/$192 yearly; Team $60/$648 yearly; Ultra $200 monthly.
- Fixed local/Pro examples: one Free user $0; two Pro users $40/month or $384/year; five $100/month or $960/year.
- Superset became Paradev on October 26, 2026; latest stable desktop-v2.0.5.
- Paradev states it completed a SOC 2 Type II audit (Aug 27, 2026 changelog entry) and lists the SOC 2 Type II report…
- Free covers one local desktop/CLI user without a card.
- Seat prices, pooled Cloud GB-hours and $0.06/GB-hour extras are public.
Sources
Open the Paradev page
T3 Codefreshness recheckresearch correction48 findings · 40 sources
Findings today
- v0.1.7 documents Mac/Windows/Linux desktop, iOS/Android mobile, web and CLI.
- v0.1.7 documents installed Codex, Claude Code, Cursor, Grok Build, Kiro (beta, Early Access), OpenCode, Antigravity,…
- T3 Code reuses each provider's own subscription or login, supports multiple accounts per provider and per-instance…
- First-task setup needs an installed, signed-in provider on the server machine; projects can start without a repo or…
- The t3 CLI and remote HTTP MCP expose project/thread operations in v0.1.7.
- T3 Code relies on the underlying harness for MCP servers, plugins and skills; it reads skills from harness config…
- Source inspection at v0.0.45 shows an Electron desktop client and an Expo/React Native mobile client alongside a web…
- Many threads can run at once, including background starts and multi-model fan-out into separate worktrees; sidebar…
- Stable v0.1.7 has T3-owned MCP delegation across compatible providers, with role/model selection, scoped…
- Users can queue follow-ups or steer new messages into a running turn, interrupt the turn with the thread.stop…
- The sidebar groups and searches saved threads; Agents views and cards expose delegated children and retained results…
- The public repository is MIT-licensed (T3 Tools Inc) and includes desktop, mobile, web, server and the T3 Connect…
- Self-hosted environment servers can use LAN, SSH or optional T3 Connect.
- The live privacy policy describes anonymous usage data sent by default; the pinned telemetry guide describes a…
- Terms state no T3 Tools subscription fee currently, unless a feature states otherwise; they do not mention seats or…
- The current terms state no T3 Tools subscription fee and do not mention seats or user counts; provider usage and…
- Latest stable is v0.1.7, displayed November 4; v0.1.8 nightly builds are separate.
- T3 Code is currently offered without a T3 Tools subscription fee, and the terms allow paid features with prices…
- The pricing URL returned HTTP 404 and the homepage was reachable in the November 7 second pass.
- Tasks start as prompts in a thread; pull requests can be referenced as context with #.
- Plan mode and proposed-plan artifacts exist in current stable.
- Skills are discoverable per provider and project; shared memory or provenance features were not documented.
- Image, PDF, text and audio attachments are documented; code search, indexing and API-awareness features are not…
- Agents run on the user's own machine or a machine they keep running (SSH, T3 Connect, LAN/tailnet); clients on other…
- T3 Code uses project checkouts or per-thread Git worktrees and launches provider/ACP processes on the environment host.
- Threads persist on the environment server.
- Project actions and automatic submodule initialization for new worktrees are documented; desktop-managed SSH relies…
- Provider usage-limit stops (for example Claude subscription limits, and Codex windows when Codex reports them) are…
- The Usage page reports token use, cache savings, model breakdowns and estimated cost over 24 hours to 90 days, with…
- v0.1.7 Usage shows provider/token estimates and model trend, cache-hit-rate and cost-per-million drilldown.
- Optional client-side load balancing (Auto balance, off by default) picks a machine for new threads using available…
- Workspace file changes are captured in hidden Git checkpoint refs, and the Terms describe T3 displaying file changes…
- Threads can use per-thread worktrees and branches, commit, push and open pull requests with generated titles and…
- Merging a pull request from the Pull requests page is documented, including auto-merge (GitHub, GitLab, Azure…
- Linked pull requests, PR status badges and GitHub revert-PR creation are documented; rollback via checkpoints and…
- Optional beta PR review guides provide ordered stops, grouped mechanical changes, retained review context and agent…
- Thread permission modes expose command, edit and child approvals in the conversation, with provider-specific…
- Threads retain conversations and tool activity, but supported rewind can remove later conversation history.
- Current stable PR review includes mobile diffs/draft comments and optional beta model-written guides with ordered…
- Stable v0.1.7 runs server-owned recurring/one-time schedules and webhook tasks with signature, filtering,…
- Threads can start without a project and choose model or permission defaults.
- Partly established for T3 Code: first-party policy pages checked 2026-10-09 name no SOC 2 or ISO certification or…
- Partly established for T3 Code v0.1.7 at thread level: threads can exist without a project, and settling keeps…
- T3 Code has persistent threads and scheduled automations, but the scoped schema/contracts have no separate native…
- Partly established for T3 Code v0.1.7: pull requests can be reviewed, reviewers requested and merged from T3;…
- Optional beta guestlinks share recent/live thread history and accept guest messages/files, gated by the owner…
- T3 Code controls devices with environment pairing credentials, session tokens and scoped grants; administrators can…
- Portable provider handoff preserves a bounded selection of saved conversation context and retrieval references.
Sources
Open the T3 Code page
8 ADEs
Google Antigravityinitial researchresearch correctionfreshness recheck43 findings · 48 sources
Findings today
- One Antigravity harness with plan-dependent Gemini, Claude and GPT-OSS models.
- Free, Google AI Pro/Ultra subscriptions, AI credits and Gemini Enterprise licenses are documented; API keys work in…
- Desktop onboarding works with plain local folders or Git checkouts; hosted Git setup is not documented; a preview…
- MCP client, skills, plugins and hooks are supported; AGENTS.md and GEMINI.md are read, CLAUDE.md import is not…
- Work starts from prompts in a local project conversation; external tracker intake was not found.
- /plan produces a reviewable plan artifact with inline comments and agent revision; persistence and revision history…
- Teamwork breaks a brief into milestones and parallel tracks; user-visible epics, blockers and readiness are not…
- Rules, AGENTS.md, skills and a /learn command provide shared instructions; memory behavior is not documented.
- Workspace code search, URL reading tools, MCP servers and image/Office inputs are documented; indexing and citation…
- Parent agents invoke subagents and Teamwork routes work, assigning each file to one worker; human assignment and…
- Custom subagents and Teamwork roles support messaging, result return, live progress and follow-up; cross-harness…
- Steering, approval cards, resume, notifications and subagent stop are documented; pause semantics and error…
- Quick tasks, plan mode, Teamwork for large work and natural-language custom agents are documented; run-time role…
- Local conversations can use project checkouts or Git worktrees, and the terminal sandbox uses native OS isolation.
- Sessions can be resumed, the Remote Control web UI tries to reconnect after network drops, and the headless Remote…
- Remote host setup via a headless daemon is documented; dependency and cache handling are not.
- Headless print mode defaults to unlimited time with an optional --print-timeout.
- Headless status values, approval cards, enterprise cloud telemetry logging, Remote Control daemon log locations and…
- Per-model quota usage, headless token usage and credit balance alerts are documented; per-task cost is not.
- Enterprise SSO, Workforce Identity Federation and BYOID are documented; licenses are assigned by the organization's…
- Command, MCP, URL and file approvals with allow/ask/deny rules and presets exist; approver identity and revocation…
- Conversation histories persist across sessions and Teamwork records logs; export and integrity guarantees are not…
- Inline comments on plans and diffs, a Review Changes pane, walkthrough summaries and keyboard diff navigation are…
- Session history and cross-device access to the same session exist for the same account; colleague handoff is not…
- Agents run tests, hooks can run linters and block stops, and Teamwork adds reviewer gates; configured CI gates are…
- Worktrees, a Git-native VCS panel and local commits with generated messages are documented; pull-request features…
- Firebase skills can deploy security rules; merge and release authority are not documented.
- /rewind rolls back conversation history; code revert semantics and delivery links are not documented.
- The main Google Antigravity service and downloadable application use Google-authored service/software terms, with…
- Desktop threads can be imported into the CLI; 2.0 v2.19.1 can export rendered Markdown in artifacts as PDF; SDK…
- Individual use lets Google use Interactions to improve products with a settings opt-out; enterprise data is…
- Free plan $0/month; Business Standard/Plus from $30 USD per seat per month for eligible Gemini Enterprise customers,…
- App v2.29.0 dated November 3 and CLI v1.5.1 dated November 5 are listed.
- Google Cloud documentation states Antigravity in Gemini Enterprise does not support ISO or SOC 1/2/3 certifications.
- Pricing units are per seat per month or consumption; included limits are qualitative; subscription options require…
- Desktop on macOS 12+ with Apple Silicon or Intel, Windows 10 64-bit+, Linux and Googlebook; CLI and IDE extensions.
- CLI headless mode, a Python SDK and a preview managed agent API exist; webhooks and a hosted MCP server were not found.
- Antigravity 2.0 runs multiple local agents across projects and workspaces in parallel, with worktree mode and…
- Automations (formerly Scheduled Tasks) and /schedule run prompts once or on a cron schedule; where they run and…
- Remote Control lists multiple machines and sessions with push attention alerts; 2.0 shows subagents in a panel and…
- Local execution, browser Remote Control of user hosts, and gradually rolling out Google-managed Remote Agents on…
- Diffs, plans, walkthroughs, diagrams, screenshots and browser recordings are produced as artifacts.
- Individual plan is $0/month with basic weekly rate limits and unlimited Tab completions; Pro and Ultra add limits…
Sources
Open the Google Antigravity page
Claude Codeinitial researchresearch correction45 findings · 62 sources
Findings today
- Claude Code embeds its own harness and runs Claude models only.
- CLI onboarding is one install command plus browser login, and works in any directory.
- Programmable access is strong on the CLI: claude -p with JSON output, Python/TypeScript Agent SDK, claude mcp serve…
- Claude Code is an MCP client (http, sse, stdio, websocket) with OAuth and scoped config, supports…
- The mobile app is a client for cloud or Remote Control sessions, and IDE extensions wrap or bundle the CLI engine.
- No native issue tracker was found in the documentation, changelog or repository checked; work arrives by prompt,…
- Plan mode proposes a plan before edits and plans can be saved into the repository; /goal tracks a completion condition.
- Ordinary TaskUpdate changes task dependencies.
- Code understanding relies on agentic search (grep, glob, file reads) plus optional LSP code-intelligence plugins and…
- Living product specifications were not found as a native feature; instructions files and plans in the repository are…
- Parallel work is supported locally (multiple terminals, worktrees, subagents, agent view background sessions, Remote…
- In the CLI a human can interrupt, stop, queue steering messages and resume sessions; background sessions show Needs…
- Agent view (research preview) is a terminal fleet screen that groups background sessions under Needs input, Ready…
- Quick tasks are one prompt away; larger efforts use subagents, agent teams, projects (cloud beta) or dynamic…
- Claude Code runs locally by default; cloud sessions run on Anthropic VMs (4 vCPU, 16 GB, 30 GB) with no separate…
- Isolation options: git worktrees per session, an OS-level sandbox for shell commands (macOS/Linux/WSL2, off by…
- Sessions persist as local transcripts and resume; cloud sessions restore history after VM reclamation but lose…
- Cloud environments offer setup scripts, a filesystem-snapshot cache, network access levels and pre-installed…
- Limits include print-mode --max-turns and --max-budget-usd (client-side estimate), subagent maxTurns, tool timeouts,…
- Operational statistics are available via /usage, /insights, OpenTelemetry metrics, logs and traces (beta), and the…
- Cost views: /usage session cost (client-side estimate at list price unless an administrator's managed modelPricing…
- Compute-aware scheduling is limited: Remote Control and self-hosted runners cap concurrency, Bash commands have…
- Live multi-human collaboration is limited.
- Roles and access are managed at the Claude account level: Owners and admins configure cloud environments and…
- Approvals are per-user permission prompts governed by permission modes plus allow/ask/deny rules, hooks and…
- History comes from local session transcripts, checkpoints, OpenTelemetry events and, on Enterprise, audit logs.
- Review is available through /diff, Desktop diff view with inline comments, local /code-review, and managed GitHub PR…
- Takeover is mostly same-user: teleport and Remote Control move a session between one person own devices; colleagues…
- Outputs include diffs (/diff, Desktop diff view), live Artifact pages published to claude.ai, Desktop app preview,…
- Verification is policy you configure: hooks run commands and can block actions or task completion (exit code 2),…
- Merge and deployment authority is delegated to git/CLI permissions: Bash allow, ask or deny rules can gate push and…
- Completion evidence is session links to PRs and OTel commit/PR counts; rollback is via /rewind checkpoints on local…
- High-volume shipping features are limited: agent view summaries, /batch splitting a change into worktree subagents,…
- Claude Code is proprietary: the public repository LICENSE.md states "© Anthropic PBC.
- The installed Claude Code CLI can run its full agent loop, tools and context management on user-controlled…
- Prompts and outputs go to Anthropic (or the chosen provider) over TLS 1.2+; consumer plans train on data only when…
- Pro $20/month or $200/year; Max $100/month; Ultra $200/month.
- Fixed subscription examples: one Pro $20/month or $200/year; two Team Standard seats $50/month or $480/year; five…
- Current changelog lists v2.1.319.
- Claude Code documentation points to the Anthropic Trust Center for compliance artifacts.
- Free excludes Claude Code; all paid plans include it.
- Public seat prices, plan multiples and separate API rates are stated.
- Persistent instructions and memory are documented: CLAUDE.md at managed/user/project/local scope, AGENTS.md reading,…
- Agent teams (experimental) document lead assignment and file-locked task claiming; outside teams, assignment is the…
- Delegation is inspectable through subagents (own model, effort, tools), agent teams, cross-session messaging and…
Sources
Open the Claude Code page
Conductorfreshness recheckresearch correctionproduct changeinitial research44 findings · 44 sources
Findings today
- Conductor documents a macOS app and an iPhone app requiring iOS 26.0+.
- Four agent harnesses (Claude Code, Codex, Cursor, OpenCode) with several model-provider routes; Claude Code, Codex…
- Subscriptions and API keys are accepted locally and in Cloud; Sign in with ChatGPT was added for Codex; Conductor…
- Git-backed setup with credential checks; GitHub required; Cloud onboarding is guided and BYO cloud machines are…
- Beta HTTP API (/v0, OpenAPI), hosted Streamable HTTP MCP server and routine webhooks; Conductor CLI exists in cloud…
- MCP client use through each harness with user/project scopes, OAuth and approval prompts; instruction files and…
- GitHub and Linear issues, branches and PRs can start workspaces; prompt links prefill the composer.
- Plan Mode for Claude and Codex and workspace todos organize longer work.
- Repository settings, instruction files and skills carry shared guidance; persistent memory is not documented.
- Attachments of notes, specs, screenshots and logs are supported, and file-content search exists (0.87.0); indexing,…
- Workspace todos exist; an independent issue backlog is unverified.
- Agents can receive workspace todos; shared native issue operations remain unverified.
- Multiple workspaces, or several chats in one workspace, run in parallel; Home and the sidebar show workspace status,…
- Cloud workspaces can be reassigned to teammates; agent assignment is not documented.
- API/MCP (beta, cloud workspaces) can create sessions, send prompts, cancel the current turn and read transcripts; an…
- Cloud routines run saved prompts on schedules (including weekdays), GitHub actions with filters, or webhooks;…
- Cloud scope.
- Home filters and sidebar sections organize workspaces; a matrix-style status indicator was introduced as…
- Managed Cloud (Vercel sandbox, us-east-1) keeps agents running after the Mac app closes until sandbox sleep or…
- Local workspaces use a git worktree and branch each without an OS sandbox; Cloud uses isolated Linux microVM sandboxes.
- Cloud preserves files and chat history across sleep but not running processes; interrupted turns must be resent.
- Cloud Computer builds seed new environments; setup scripts run per workspace; port forwarding is off by default for…
- Cloud sandboxes enforce sleep and lifetime limits; Codex chats retry on OpenAI rate limits (0.84.0, no surface named).
- Experimental cloud cost estimates (Codex, 0.77.0) exist; per-task cost reporting is otherwise not documented;…
- Shared Cloud workspaces support live chats, presence and typing indicators; private workspaces exist; Pro…
- Cloud organizations have admins and members.
- Local tool approvals exist, with settings for defaults, plan mode and managed enforcement.
- Diff viewer, line comments, GitHub review threads and suggested PR actions guide review.
- Teammates can continue and reassign a Cloud workspace.
- Cloud workspaces show diffs; plans, screenshots and attachments in .context sync from cloud workspaces to the Mac…
- Checks aggregate CI; unresolved todos block merge; the checks page says Conductor "may block or discourage" merge…
- Workspace branch-to-PR workflow is documented, with GitHub stacks (0.80.0; local needs the gh stack extension) and…
- One-click PR approval and merge from Conductor; merge blocked if the PR changed on GitHub.
- Local checkpoints support destructive turn reversion; the Checks tab may include pull request metadata and deployments.
- Conductor terms explicitly define the client-side and hosted Platform as proprietary software.
- The local app keeps chats and files on the Mac (~/Library/Application Support/com.conductor.app and ~/conductor),…
- Local workspaces and chats stay on the Mac, model requests go to the configured provider, and analytics go to PostHog.
- The pricing URL redirects to the homepage without a price table.
- Conductor has withdrawn its wind-down plan.
- Conductor reports SOC 2 Type II; audit report and scope not inspected.
- Pricing is partially specified: model billing stays with the provider and Cloud FAQ states no additional compute fee…
- Per-session harness, model, Fast Mode and reasoning-level controls exist; workflow presets, coordinator and…
- API session status distinguishes idle, working and errored; failed Cloud startup exposes setup logs.
- One-, two- and five-person scenarios cannot be priced from the checked public pages because current subscription…
Sources
Open the Conductor page
Cursorinitial researchfreshness recheckproduct changeresearch correction47 findings · 53 sources
Findings today
- Cursor runs its own coding-agent loop with selectable models.
- A Cursor account uses the new X Subscription or a retained legacy Cursor plan.
- Cloud Agents need a paid plan and connected source control (or a no-repo start); setup is agent-led or…
- Cursor exposes a CLI with headless mode, a public-beta Cloud Agents REST API, a TypeScript SDK, and signed status…
- Cursor is an MCP client with OAuth and admin allowlists, supports plugins and marketplaces, and reads skills,…
- Plan Mode produces editable markdown plans before building; acceptance criteria, revision history and plan-to-work…
- Projects let a coordinator agent plan larger bodies of work and delegate to cloud agents; task dependencies,…
- Rules, AGENTS.md, skills, team rules, automation memories and Project shared-context files carry instructions and…
- Agents search code with a local grep index, web search and MCP, and accept image inputs; embeddings-based indexing,…
- Agents interact with external Linear and Jira issues (assigned or mentioned) and report progress there; a native…
- Many agents can run in parallel across local worktrees, remote SSH machines, self-managed machines and the cloud; no…
- Work is routed by Slack keyword rules and by Jira repository-name, service-name and keyword matching; Team Pool…
- Subagents (built-in and custom) and Project coordinators delegate work with isolated context and return final…
- Automations run on schedules, source-control events, Slack, Linear, Sentry, PagerDuty and webhooks; Projects can…
- Users can steer, queue, cancel, resume and follow agents from several clients; notifications reach mobile and Slack.
- The Agents Window and the mobile app list agents; Projects show the agents their coordinator started.
- Cursor supports one-off tasks in Agent mode, longer coordinator-led Projects, Custom Modes with skills, and subagent…
- Local agents use worktrees and an OS sandbox; cloud agents run in per-agent Firecracker microVMs with configurable…
- Cloud agents accept follow-up runs and CLI sessions can be resumed.
- Spend limits bound cloud usage; time limits and timeouts are not stated.
- Logs, run states, per-run token usage and admin analytics exist; traces and approval-state reporting are not documented.
- Token usage is available per agent and per run through the API and SDK; billed dollar cost per agent is available…
- Team Pools queue requests for workers and a controller scales machines; other compute-aware scheduling and test…
- Teams have Member, Admin and Unpaid Admin roles with SAML SSO; Enterprise adds SCIM, organization groups and MDM…
- Run modes, sandbox rules, MCP approvals and hooks control agent actions locally; cloud agents run commands without…
- Enterprise audit logs record admin and configuration events with CSV, API and SIEM export; conversations persist…
- Agent Review, Bugbot, a diffs view and mobile PR review support reviewing agent output; shipping guidance and risk…
- Independent AI review (Agent Review, Bugbot) and policy files exist, and Bugbot can fail a status check; lint, type…
- Cloud agents push branches, make signed commits, open draft PRs and respond to PR comments; local work uses worktrees.
- Agents open pull requests (ready for review unless a draft is requested).
- PR, deployment and check status appear in mobile, and Rollouts monitors deploys for regressions and notifies the…
- High-volume batch review views were not found; mobile and PR routing give partial support.
- The Terms grant a limited right to access and use the Service and reserve all rights not granted; the public…
- Cursor is a vendor-hosted service; only tool execution can be self-hosted, and cancellation reverts the account to…
- Privacy Mode prevents training.
- Cursor announced its completed acquisition by SpaceX on August 14, 2026 and described continued product development.
- Cursor's pricing footer displays "SOC 2 | ISO27001 | ISO42001 | AIUC-1 Certified" (vendor statement, pricing page…
- New-plan seat rates and annual prices are public; usage pools are weekly and third-party models prepaid.
- Work can start from prompts in the app, CLI or API, and from Slack, GitHub, Bitbucket, Linear and Jira mentions or…
- Work can run on the local machine, on remote SSH hosts, on Cursor-managed VMs, or on self-managed workers; managed…
- Cloud setup builds reproducible environments from agent-led setup, snapshots or Dockerfiles with install/start…
- Teammates can continue shared cloud runs when an admin enables team follow-ups, take over the agent desktop, or fork…
- Cursor is reachable from a desktop app (Mac, Windows, Linux), a CLI, a web client at cursor.com/agents, a native…
- Agents are visible to members of the team they were started under and are read-only for viewers; follow-ups are…
- Cloud runs produce diffs, PRs and artifacts (screenshots, videos, log references) with remote desktop takeover and…
- New subscriptions cost $8 Plus, $30 Premium, $100 Super or $200 Ultra monthly; Teams seats $30/$100/$200.
- Free has limited core use.
Sources
Open the Cursor page
JetBrains Airresearch correctionfreshness recheckproduct changeinitial research40 findings · 45 sources
Findings today
- First local task needs Git, the Air app and one connected provider; cloud tasks need an organization set up in…
- Air supports MCP in the Air app (global scope available in all projects on the machine, local scope in .air/mcp.json…
- Automations can be paused and resumed and a run in progress can be stopped; suspended cloud environments and…
- Air is proprietary: the EAP User Agreement (v1.0) section 3.3 prohibits reverse engineering, decompiling, trying to…
- The EAP agreement says JetBrains does not see or access user Data in the product by default without consent and that…
- Air is in early access: the Air app is under an EAP user agreement, Air in IDEs is EAP, Air Teams is available to…
- Task context can reference files, folders, symbols, Git branches and commits, local changes, terminals and uploaded…
- The Air app offers Plan, Ask, Edit and Full Access permission modes; Ask requires approval before changes or…
- The Air app shows a merged Task Changes diff with unified or side-by-side views, line comments and Add to Task…
- Air is reachable through the Air app on macOS, Linux and Windows, Air Teams in the browser, and Air in JetBrains…
- Air in IDEs runs local coding agents through ACP and terminal surfaces, explicitly including Claude Code, Codex and…
- Air Teams automations can be started by an incoming webhook and can send outgoing webhook notifications; the Central…
- The Air app runs several tasks at once across projects, Air Teams runs each cloud task in its own environment, and…
- A review pass can run in a separate agent session with a different agent and model and its accepted comments are…
- Air Teams automations run in the cloud on a schedule, at an interval, or from GitHub, Jira or webhook events without…
- Air app tasks group by date, status or target and flag Input required.
- The Air app can run local tasks without Air Teams or an organization on the user's own provider account.
- Console Business is $20 per seat per month (monthly or annual billing); 1 AI credit equals 1 USD of provider-side…
- JetBrains states SOC 2 Type II covers the Central Console and its governance layer and that all plans are GDPR…
- Air app EAP terms allow free app use; AI usage is separate.
- Units are AI credits (1 credit per 1 USD of provider-side LLM cost) and Business plan AI seats at $20 per seat per…
- Tasks are created from a prompt in the Air app or browser; automations can start from GitHub, Jira or webhook events.
- Plan mode lets the agent write a plan as a Markdown file in .air/plans before implementation and the user approves…
- Air in IDEs runs agent sessions on the user's machine by default, with Local and Temporary Worktree modes, and Air…
- In the Air app, agentic review runs in a separate agent session; the browser has no built-in review pass.
- Air Teams projects share environments, connectors and automation results.
- Air Teams uses organization membership and project admin/member roles.
- Central Console keeps an immutable audit log of organization administration changes (not AI activity) with CSV…
- Air picks up CLAUDE.md and the .claude folder for Claude Agent, reads AGENTS.md for agent workflows, and loads…
- A task starts from text typed in the task input.
- Run environment is selectable: Local Workspace (no isolation), Git Worktree, Docker (local) and Cloud; worktree and…
- Cloud tasks keep running after the browser tab closes.
- Cloud environments use a startup script, snapshots and shared configurations; worktree and Docker tasks use…
- Documented limits are the monthly credit limit, a startup script timeout (default 600 s, 1-3600 s) and inactivity…
- Cloud tasks expose a Problem status with the error at the top of the task page, downloadable logs (Collect Logs and…
- In the documented current release, Air Teams usage tracking covers JetBrains AI only, and the Air app credit counter…
- Diffs are reviewed in the Task Changes tab and their commits remain on the task branch; plans are Markdown files in…
- Cloud tasks commit to a task branch and push after every agent turn; a person then opens the pull request from that…
- Before a commit, the Task Changes tab and the Changes tool revert changes file by file or all at once.
- The Comments tool can show comments for the current task or across tasks, diff stats summarize size, and Tasks tool…
Sources
Open the JetBrains Air page
Multicainitial researchresearch correctionfreshness recheckproduct change47 findings · 58 sources
Findings today
- The vendor website, the providers table and the install table each list 29 supported coding tools, including…
- Multica does not provide model access in the sources read: each coding tool uses the account, subscription or API…
- Documented onboarding is: sign in, connect a computer with Desktop or multica setup, create an agent, assign the…
- Workspace owners/admins keep an MCP server library that is assigned per agent and passed to most supported tools;…
- Multica issues are the native record: issues are created in the app, CLI, chat bots or by prompting an agent.
- Issues hold the task brief and requirements (description text, comments, attachments); no separate structured plan…
- Agent instructions, workspace context, project descriptions and reusable skills provide shared context; sessions…
- Multica passes attachments and chat media to agents but no product-level code search, indexing, architecture…
- Multiple runs execute concurrently on a daemon (default 20 per daemon and 6 per agent) and work state is visible in…
- Squads provide product-owned delegation: an agent leader receives the issue, @-mentions member agents, records an…
- Autopilots run on 5-field cron schedules or webhook triggers from the Multica server, and issue wakeup rules can…
- Issues can be browsed in list, board, table, Gantt and swimlane views with status, assignee and project filters,…
- No mandatory workflow preset is required; a quick chat or issue works alone, work can grow into sub-issues, and an…
- Connected computers execute local-tool runs while Cloud or self-hosted Multica coordinates them.
- GitHub repository runs get their own worktree from a cached bare clone; local git directories can use per-run…
- Repository checkouts use cached bare clones and per-task worktrees; the user supplies and maintains tool installs…
- The daemon enforces a configurable absolute run time limit (default none), an idle watchdog and a tool watchdog…
- A daemon-wide cap (20) and per-agent cap (6) queue excess runs, and a directory lock serializes runs on the same…
- Workspace members share issues and comments with attribution; members can each comment, mention and steer agents…
- Workspaces have owner, admin and member roles, 7-day invitations and immediate revocation on removal; agent Access…
- Runs are unsandboxed and tool approval prompts are answered automatically by default; agent Access controls who can…
- The issue timeline records attributed activity and comments with actor type and id, comment revisions and resolve…
- Review is issue- and PR-centric: in_review status, a pull request card with state, size, CI and mergeability,…
- Issues, comments, transcripts and attachments live on the server, so another member can follow the same issue and…
- Issues retain comments, attachments with versions, a deliverables sidebar and in-place previews (HTML, Markdown,…
- Multica shows CI status and mergeability on the PR card and can move an issue when linked PRs merge, but no…
- Agents run Git and the GitHub CLI themselves; Multica adds per-task worktrees, a task Git identity, optional…
- Merging is done by humans or by agents on request through the user's own Git tooling; the GitHub App is read-only.
- Linked pull requests show state and merge status on the issue and merged PRs can complete it; no build, deployment,…
- Review happens one issue at a time through the inbox, in_review status and PR cards; batch queues, grouping, risk…
- Workspace content (issues, comments, run records, attachments) is stored by Multica; code and tool credentials stay…
- No public price list was found: /pricing renders a not-found page and the site offers Get started, Book a demo and…
- Current release is v0.8.1 dated November 6.
- Self-hosting is free for internal use (infrastructure and tool accounts are the user's), the website offers a Start…
- Web app, Desktop app (macOS, Windows, Linux) and a CLI exist; the iOS client is self-built from source, with no App…
- The CLI exposes broad issue, agent, project, autopilot and runtime operations over the shared server API,…
- Native issues include creation, editing, comments, priority, labels, assignment, search, filters, projects, five…
- Issues can be assigned to members, agents or squads; runs queue for a runtime and are claimed by it; consecutive…
- Run history exposes stop and retry; running Claude Code, Codex and Grok tasks can be steered with new instructions;…
- Transient failures auto-retry (default 2 attempts, 3 for tool network errors) and a daemon restart reclaims…
- Execution log exposes run state, timing and streaming transcript with classified failure reasons; a workspace…
- Multica can be self-hosted (Docker Compose, Kubernetes) with Desktop and CLI pointed at it; self-hosted use needs no…
- Desktop uses Electron; mobile docs specify React Native.
- Issues have parent-child relationships.
- Agents use CLI against shared issue records.
- Estimated per-run and per-issue cost appears in the execution log with a cumulative curve, and the CLI reports issue…
- Custom Multica License includes extra conditions.
Sources
Open the Multica page
Orcaresearch correctionfreshness recheckinitial research51 findings · 38 sources
Findings today
- Work can start locally from a name or prompt, or from GitHub, GitLab, Linear or Jira items; the external tracker…
- Review is built into each worktree: a combined diff, image and HTML preview, three-way conflict resolution, line and…
- Diffs, HTML and Markdown previews, per-worktree browser pages, screenshots through the CLI and shared HTML or…
- Orca keeps local per-line attribution of human versus AI edits, a feed of agent events, searchable on-disk session…
- The public repository carries a root MIT license that covers the desktop app, mobile app, CLI and the cloud/ Relay…
- Desktop on macOS, Windows and Linux, plus beta iOS/Android companion.
- Orca launches and tracks existing CLI agents in terminals rather than embedding a harness: many CLI launchers are…
- Orca reuses the installed coding-agent native subscription logins, including Claude Code and Codex.
- Orca exposes a broad orca CLI (worktrees, terminals, files, browser, orchestration, automations, Linear, accounts,…
- Source identifies an Electron and React desktop app and an Expo and React Native mobile app; docs mention a…
- The desktop app is advertised as free and open source, so BYO-subscription and self-hosted scenarios carry no Orca…
- Orca documents native Run/Task/Dispatch records independent of external trackers, but the required Experimental…
- Agents can read, create and update Linear issues through orca linear and set worktree comments and status, report…
- Local desktop, user-owned SSH hosts and persistent Remote Orca Servers; experimental Cloud VM recipes use the user…
- Each task is a real git worktree with its own branch and directory.
- A background daemon keeps local agent processes running across app quit, update relaunch and Orca app crashes.
- Orca shows provider usage windows, resets and credit balances, and Stats with estimated cost from a local price…
- The Enterprise page shows an "AICPA SOC 2 Readiness" label and the question "Is Orca SOC 2 compliant?", The captured…
- Documentation describes install, first launch (home-directory access, optional import of Claude, Codex and Ghostty…
- Orca registers MCP servers for MCP-capable agent CLIs, installs agent skills, imports skill folders from Claude,…
- Orca keeps lightweight task state: orchestration tasks carry a spec and status, worktrees carry a free-text comment…
- Workspaces can be nested under a parent workspace and orchestration tasks have dependencies and statuses including…
- Orca relies on the agents own instruction files (CLAUDE.md, AGENTS.md) and repo hooks, offers installable and…
- Multimodal inputs are documented: files and images dragged into agent prompts, Design Mode sends element HTML, CSS…
- Native orchestration tasks retain a spec and link to execution attempts and decision gates.
- Multiple agent sessions run concurrently, each in its own worktree and terminal, with status visible in the sidebar,…
- Orchestration dispatches a task to a specific worker terminal and treats the active dispatch as completion…
- Orca documents its own orchestration skill/CLI and supervised worker delegation, but the required Experimental panel…
- orca automations run a prompt on hourly, daily, weekday, weekly, cron or RRULE schedules against a repo or worktree,…
- Users can stop, steer, queue messages, answer pending questions and approvals (in terminal or Chat UI), and are…
- Fleet views documented: sidebar worktree cards with N-children chips, an experimental kanban dashboard (Needs You,…
- Creating a worktree and launching an agent is a create-and-launch sequence with no mandatory workflow selection (the…
- New worktrees can run setup hooks, share gitignored directories such as node_modules by symlink, and copy .env-style…
- No enforced time, turn, token or spend limit was found.
- Agent state is shown from hooks and OSC titles with a distinct waiting-on-you state, a feed and a dashboard; Stats…
- A Resource Manager shows CPU, memory and session counts with daemon controls, and idle agents can hibernate to free…
- Orca is built around one developer machine or server; several paired clients can see the same server-owned workspaces.
- The self-managed Orca runtime uses revocable paired-device grants without named organization/project user roles or…
- Approval handling is largely delegated to the agent CLIs: by default Orca pre-fills each supported CLI's…
- A second client paired to a Remote Orca Server reaches the same server-owned workspaces, tabs and agent sessions.
- Repo pre-commit hooks run on commit with an inline failure and a Fix with AI repair prompt, failing GitHub checks…
- Branches, worktrees, staging by hunk, commits, push, PR or MR creation (GitHub, GitLab, Bitbucket Cloud) with draft…
- Orca can merge or enable auto-merge on GitHub pull requests using the repository allowed merge methods and can merge…
- Worktrees link to pull requests and merge requests with live status, and CI check results appear on the worktree.
- A done queue exists in the experimental dashboard and Agents feed with response previews, PR status filters and…
- Orca can be fully self-hosted for desktop, SSH and Remote Orca Server use with no Orca account; Relay pairing,…
- Vendor policy describes local code/provider credentials and anonymous opt-out telemetry.
- The Orca application is advertised as free and open source with no price list; the Enterprise page offers rollout…
- No priced scenario was built.
- Latest GitHub release is v1.4.253, published November 7, 2026, following v1.4.252 the same day and v1.4.251 on…
- The site advertises the desktop app as free; no desktop seat, usage-unit or overage rate is established by the cited…
Sources
Open the Orca page
Paseofreshness recheckproduct changeinitial researchresearch correction46 findings · 40 sources
Findings today
- Paseo offers desktop, native phone, browser and CLI clients for its daemon; the shared browser app includes compact…
- Paseo reuses each provider CLI's own login for Claude plan, ChatGPT plan and API-key use, and adds no charge for…
- Paseo works in any directory on a daemon host, with or without git; worktree features and PR workspaces need a git…
- Paseo exposes a CLI, TypeScript SDK, daemon MCP tools and a Hub HTTP API with scoped keys and OpenAPI; daemon access…
- Paseo desktop is an Electron app; the mobile and web clients share an Expo / React Native app package.
- Paseo runs multiple agents and workspaces concurrently, each in its own worktree if wanted; documented recipes cover…
- Agents can launch, prompt, inspect and cancel other agents of any provider; workers are retained as inspectable…
- Paseo schedules start new agents on a cron cadence (with run and expiry limits) and heartbeats re-prompt an existing…
- Users and parent agents can cancel a turn, stop or kill an agent, send follow-ups, change mode, and answer…
- Paseo runs agents on operator-owned local or server daemon machines.
- Paseo isolates file changes with optional git worktrees per workspace; same-workspace agents share files.
- In self-hosted Paseo Hub, distinct permitted Slack users can send instructions to one daemon agent by enabling Same…
- Paseo v0.15.3 root code is Apache-2.0 with third-party licence exceptions; its desktop manifest also declares…
- The Paseo app, daemon, CLI and SDK are free and open source with no usage cap documented; hosted Hub offers a…
- Current stable and homepage document native Claude Code, Codex, OpenCode, Pi, OMP, Antigravity and Muse plus a…
- Paseo documents a plugin system and passes through the provider CLI's own MCP servers and skills; it does not…
- Paseo shows each subscription account's quota windows and context-window usage; per-task or per-epic dollar cost,…
- The Paseo app, daemon, CLI and SDK are free; hosted Hub is Free ($0, 50 agent runs/month, 1 seat) or Pro ($15 per…
- Paseo is workspace- and session-based and is not documented as having a native task-tracker intake; agents start…
- Paseo documents plan mode as a session mode and persists agent timelines; a durable task plan or acceptance-criteria…
- Paseo models parentage between agents (subagents, detach) and workspaces; epics, blockers and readiness states are…
- Paseo relies on each provider CLI's instructions, skills and memory; it adds installable orchestration skills and…
- Paseo documents file browsing, a built-in browser and screenshot tools; code search, indexing and retrieval are not…
- Paseo documents agent parentage and per-agent addressing; human assignment, agent claims and queue routing are not…
- Paseo has a project/workspace sidebar and Subagents track but a fleet board with dependencies and workload views was…
- Paseo offers profiles, skills and natural-language orchestration; richer workflow flexibility questions are…
- Paseo persists agent records and timelines and resumes closed agents under the same ID.
- Paseo worktree setup hooks, scripts and supervised services with port allocation address environment setup; caching,…
- Paseo documents run counts and expiry on schedules and max_runtime / idle_timeout on Hub steps; token, spend and…
- Paseo provides agent timelines (paseo logs), daemon and desktop logs, a Usage view of subscription quota windows and…
- Paseo documents no compute-aware scheduling; Hub dispatch is not queued and agents are launched on demand.
- Paseo Hub has owner, admin and member organization roles with enforced management permissions.
- Paseo routes provider permission prompts to the app, CLI and MCP; Hub runs use the provider permission modes set on…
- Paseo persists agent timelines and Hub records every accepted event and run; export, retention and integrity…
- Paseo provides a Changes (diff) tab, Files, PR panels with check summaries and split panes; a pending-review queue…
- Another user with pairing credentials can connect to the same daemon and, with workspace.read, see its workspaces…
- Paseo shows workspace diffs and files, previews apps in built-in browser tabs and service proxy URLs, and browser…
- Paseo lets users define named workspace scripts (tests, lint) and run them from app, CLI or agents; required checks,…
- Paseo creates worktrees and branches, drafts commit messages and PR text, opens PR workspaces and detects linked…
- Merge and deployment authority are not documented by Paseo; Hub steps limit GitHub authority via declared…
- Paseo documents PR links for workspaces; commit, build and deployment links, completion evidence and code rollback…
- Paseo runs entirely on user machines, with optional relay and Hub; self-hosting Hub is documented.
- Stable v0.15.3 states no telemetry, tracking or required Paseo account for the local app.
- Fixed hosted Hub examples: one Free seat $0 for 50 runs/month; two Pro seats $30/month; five $75/month.
- Current stable is v0.15.3 on November 6, with separate Beta channel and gradual desktop rollout.
- Hub pricing lists the unit (seat), included runs and the seat definition; Paseo documents no model or compute charge…
Sources
Open the Paseo page
Archived event links
Something wrong or missing?