# Happy

Podium publishes this catalog and sells an agent development environment. Podium chooses the products, criteria, sources and wording, so treat these pages as vendor-authored research. Check important claims against the linked primary sources. The same evidence rules apply to Podium and its competitors, but they do not remove this conflict of interest.

Page: https://adecomparison.com/ade/happy/
Website: https://happy.engineering/
Research: covered

Method: https://adecomparison.com/method/

Rank by must-haves met, then preferences, then name. Read each finding's checked date. Findings stay published after their scheduled recheck. Checks over six months old may be outdated.

## happy:ACC-01

Finding: Happy provides Desktop for macOS, Windows and Linux; mobile iOS, Android and web; and its original happy CLI wrapping Claude Code or Codex. Desktop uses a separate Happy Agent runtime for multi-model sessions and teams. Paired phone apps continue the same encrypted sessions and allow users to read work and respond away from the desktop. The first-party repository describes the mobile app as available on iOS, Android and web. This establishes the phone web client, not desktop parity.

Verification: sourced
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Happy Agent v2026.11.4 (db813011c038b2288fbc1efd0e10ef2149c838d2), Desktop v0.0.103 (a369b7b0a86a344904a2e171a78e2b0daf0b4be8); targeted first-party pass 2026-11-13; no installed-product test

The scoped first-party source pass does not establish editor_extensions, notifications. No hands-on scenario was run.

Source: https://github.com/slopus/happy/blob/df13fc0678ec4093b85fe97c0c48ec9422f5a7c2/README.md
Locator: How Happy fits together; Original Happy CLI
Finding: Happy provides Desktop for macOS, Windows and Linux; mobile iOS, Android and web; and its original happy CLI wrapping Claude Code or Codex. Desktop uses a separate Happy Agent runtime for multi-model sessions and teams.
Retrieved: 2026-11-08

Source: https://happy.engineering/mobile/
Locator: Mobile access
Finding: Paired phone apps continue the same encrypted sessions and allow users to read work and respond away from the desktop.
Retrieved: 2026-11-08

## happy:ACC-02

Finding: Happy Agent runs its own loop by default and calls Claude, Codex and Grok models on existing CLI sign-ins. Optional harnesses run Claude Code itself (claude_code) and Codex CLI itself (codex_cli); Gemini runs through Antigravity and OpenCode or other ACP agents through ACP (beta). Grok is a model choice, not a separate harness. Model and effort can change mid-session without discarding the transcript. The original happy CLI wraps Claude Code and Codex directly. Happy Agent v2026.11.4 runs the Antigravity whole-agent harness through antigravity-acp@1.2.0, a community ACP adapter driving the signed-in agy CLI.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08 Antigravity source pass: v2026.11.4 (db813011c038b2288fbc1efd0e10ef2149c838d2). Antigravity integration is beta.

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.
Antigravity is a beta foreign-harness integration using a community ACP adapter and its own login; new sessions are serialized to avoid conversation binding collisions.

Source: https://github.com/slopus/happy/blob/df13fc0678ec4093b85fe97c0c48ec9422f5a7c2/README.md
Locator: How Happy fits together; Original Happy CLI
Finding: Happy provides Desktop for macOS, Windows and Linux; mobile iOS, Android and web; and its original happy CLI wrapping Claude Code or Codex. Desktop uses a separate Happy Agent runtime for multi-model sessions and teams.
Retrieved: 2026-11-08

Source: https://happy.engineering/models/
Locator: Supported providers; Switch models
Finding: Happy runs Claude, Codex and Grok on the user's existing Claude Code, Codex or Grok CLI sign-ins, Gemini through Antigravity, and OpenCode or other ACP agents on their own login (beta). Model or effort can change mid-session without discarding the transcript.
Retrieved: 2026-11-08

Source: https://happy.engineering/recepies/harnesses/
Locator: Harnesses
Finding: Happy's own harness runs the loop by default; the claude_code harness runs Claude Code itself on a claude account and codex_cli does the same for Codex, while Antigravity, OpenCode and other ACP agents sign in themselves. Grok is not listed as a harness.
Retrieved: 2026-11-08

Source: https://github.com/slopus/happy-agent/blob/db813011c038b2288fbc1efd0e10ef2149c838d2/packages/happy-providers/sources/vendors/foreignAgents/antigravityAgent.ts#L30-L63
Locator: packages/happy-providers/sources/vendors/foreignAgents/antigravityAgent.ts lines 30-63, v2026.11.4, commit db813011c038b2288fbc1efd0e10ef2149c838d2
Finding: Happy Agent v2026.11.4 runs the Antigravity whole-agent harness through antigravity-acp@1.2.0, a community ACP adapter driving the signed-in agy CLI.
Retrieved: 2026-11-10

## happy:ACC-03

Finding: Happy reuses existing Claude Code, Codex and Grok CLI sign-ins and their subscriptions; Antigravity and ACP agents use their own login. Bedrock API keys or the AWS credential chain are supported.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/models/
Locator: Models and Subscriptions
Finding: Happy uses existing Claude Code, Codex and Grok CLI sign-ins, so no Happy plan or pasted key is needed for model access; Antigravity and ACP agents use their own login. Bedrock works with an API key or the AWS credential chain.
Retrieved: 2026-11-08

## happy:ACC-04

Finding: Importing registers an existing folder without moving, copying or initializing Git. A project without Git uses a plain folder copy when a workspace is needed.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/workspaces/
Locator: Add a project
Finding: Importing registers an existing folder without moving, copying or initializing Git. A project without Git uses a plain folder copy when a workspace is needed.
Retrieved: 2026-11-08

## happy:ACC-05

Finding: Happy Agent exposes a token-authenticated REST API for sessions, streaming events and control on user infrastructure. The original happy CLI is a separate product-owned terminal entry point.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/recepies/rest-api/
Locator: REST API
Finding: Happy Agent exposes a token-authenticated REST API for sessions, streaming events and control on user infrastructure. The original happy CLI is a separate product-owned terminal entry point.
Retrieved: 2026-11-08

## happy:ACC-06

Finding: Skills are SKILL.md files and MCP servers are mcp.toml entries; agents can write both. MCP tools need Auto or Full access, every Auto call is reviewed, and MCP servers run outside the OS sandbox.

Verification: sourced
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/extending/
Locator: Skills and MCP
Finding: Skills are SKILL.md files and MCP servers are mcp.toml entries, shared by every model; an agent can write both. MCP tools need Auto or Full access and every Auto call is reviewed. MCP servers from Codex or Claude Code configuration are not imported.
Retrieved: 2026-11-08

Source: https://happy.engineering/permissions/
Locator: What the OS sandbox enforces
Finding: The documented OS sandbox uses macOS Seatbelt or Linux namespaces and seccomp. Shell writes, network and host sockets are restricted; file tools can still read credential folders, and MCP servers run outside the sandbox. Docker-backed sessions nest that sandbox inside the container.
Retrieved: 2026-11-08

## happy:PLAN-01

Finding: The built-in Chief of Staff handles setup and task creation through ordinary conversation, including projects, remote hosts and teams. Persistent standalone issue metadata is not established.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/chief-of-staff/
Locator: Chief of Staff
Finding: The built-in Chief of Staff handles setup and task creation through ordinary conversation, including projects, remote hosts and teams. Persistent standalone issue metadata is not established.
Retrieved: 2026-11-08

## happy:PLAN-03

Finding: Subagent trees are capped at depth three and at most five agents including finished children. Sessions have exact agent IDs for messaging and retained follow-up. This is session hierarchy, not verified issue dependency enforcement.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/agents/
Locator: Subagents
Finding: Subagent trees are capped at depth three and at most five agents including finished children. Sessions have exact agent IDs for messaging and retained follow-up. This is session hierarchy, not verified issue dependency enforcement.
Retrieved: 2026-11-08

## happy:PLAN-04

Finding: AGENTS.md files in the project tree are read by each agent and reread when they change; global AGENTS.md applies across projects.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/workspaces/
Locator: Add a project
Finding: AGENTS.md files in the project tree are read by each agent and reread when they change; global AGENTS.md applies across projects.
Retrieved: 2026-11-08

## happy:PLAN-06

Finding: Happy creates persistent task records through its API, agent tool and Desktop UI. Native task operations rename task records. Creating a task without opening text stores an idle draft. The task catalog has a task ID separate from its linked agent and workspace IDs; each task has one root conversation. Task state is persisted as active or archived with restoration and versioned updates. Task archival persists and stops the task agent and descendants. Native unarchive restores an archived task.

Verification: sourced
Checked: 2026-11-13
Scheduled recheck: 2027-03-13
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Happy Agent v2026.11.4 (db813011c038b2288fbc1efd0e10ef2149c838d2), Desktop v0.0.103 (a369b7b0a86a344904a2e171a78e2b0daf0b4be8); targeted first-party pass 2026-11-13; no installed-product test

The scoped first-party source pass does not establish discuss, prioritize, label, assign, search, filter, project_epic_views, explicit_dispatch, metadata, external_tracker_requirement. No hands-on scenario was run.

Source: https://github.com/slopus/happy-agent/blob/db813011c038b2288fbc1efd0e10ef2149c838d2/packages/happy-agent-modules/sources/tasks/Task.ts#L28-L151
Locator: Task record schema and createTaskInputSchema, lines 28-151
Finding: Task records have independent task, agent and workspace IDs, owner and creator fields, versioned active/archived state and timestamps. Creation accepts no opening text, allowing an idle draft before work starts.
Retrieved: 2026-11-13

Source: https://github.com/slopus/happy-agent/blob/db813011c038b2288fbc1efd0e10ef2149c838d2/packages/happy-agent-modules/sources/tasks/TasksModule.ts#L664-L725
Locator: createWithResult and task archive/unarchive operations; corresponding TaskMigrations.ts table
Finding: The TasksModule creates distinct task and agent records, supports rename, join, reorder, archive and unarchive, and persists catalog state. The API and Desktop task creation surface expose the task catalog.
Retrieved: 2026-11-13

## happy:COORD-01

Finding: Subagents run in the background by default and return completion to the parent. Desktop displays subagents and running tools; tree size is bounded.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/agents/
Locator: Subagents; What Happy Desktop shows
Finding: Subagents run in the background by default and return completion to the parent. Desktop displays subagents and running tools; tree size is bounded.
Retrieved: 2026-11-08

## happy:COORD-02

Finding: Bots and their subtasks may create a full conversation in a new workspace, with the creator as coordinator. Only the creating session may archive that workspace.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/workspaces/
Locator: How agents work in another workspace
Finding: Bots and their subtasks may create a full conversation in a new workspace, with the creator as coordinator. Only the creating session may archive that workspace.
Retrieved: 2026-11-08

## happy:COORD-03

Finding: Agents delegate bounded work, choose each child model and effort, receive background completion notices and send follow-ups to finished children with retained context. Exact IDs route messages across projects or machines.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/agents/
Locator: Subagents; How agents message each other
Finding: Agents delegate bounded work, choose each child model and effort, receive background completion notices and send follow-ups to finished children with retained context. Exact IDs route messages across projects or machines.
Retrieved: 2026-11-08

## happy:COORD-04

Finding: Primary sessions can schedule messages to known agent IDs, list or cancel them and survive intervening runtime restarts. Durable waits end early on new input. This establishes scheduled work, not an unrestricted external-event scheduler.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/agents/
Locator: Durable waits and scheduled messages
Finding: Primary sessions can schedule messages to known agent IDs, list or cancel them and survive intervening runtime restarts. Durable waits end early on new input. This establishes scheduled work, not an unrestricted external-event scheduler.
Retrieved: 2026-11-08

## happy:COORD-05

Finding: Agent questions remain in the conversation and experimental Inbox; answering on one device closes the shared state. Away presence instructs agents to continue with notes instead of waiting indefinitely.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/agents/
Locator: Inbox and presence
Finding: Agent questions remain in the conversation and experimental Inbox; answering on one device closes the shared state. Away presence instructs agents to continue with notes instead of waiting indefinitely.
Retrieved: 2026-11-08

## happy:COORD-06

Finding: Per-session UI shows phase, tools, background processes, subagents, waiting state, whole-tree token usage and changed files. Fleet navigation has not been measured.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/agents/
Locator: What Happy Desktop shows
Finding: Per-session UI shows phase, tools, background processes, subagents, waiting state, whole-tree token usage and changed files. Fleet navigation has not been measured.
Retrieved: 2026-11-08

## happy:COORD-07

Finding: The built-in Chief of Staff configures accounts, imports projects, pairs phones, deploys remote agents and creates teams through conversation.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/welcome/
Locator: What you get
Finding: The built-in Chief of Staff configures accounts, imports projects, pairs phones, deploys remote agents and creates teams through conversation.
Retrieved: 2026-11-08

## happy:RUN-01

Finding: Happy Agent can run on an always-on user server while Desktop connects over Tailcat and phones connect through an encrypted relay. Closing the controlling laptop does not stop work on that reachable host. Users supply host compute.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/guides/remote-server/
Locator: Remote server
Finding: Happy Agent can run on an always-on user server while Desktop connects over Tailcat and phones connect through an encrypted relay. Closing the controlling laptop does not stop work on that reachable host. Users supply host compute.
Retrieved: 2026-11-08

## happy:RUN-02

Finding: Parallel workspaces use Git worktrees on branches cut from the remote trunk. File sync and setup commands run per workspace; nongit projects use a plain copy. The documented OS sandbox uses macOS Seatbelt or Linux namespaces and seccomp. Shell writes, network and host sockets are restricted; file tools can still read credential folders, and MCP servers run outside the sandbox. Docker-backed sessions nest that sandbox inside the container. No VM or microVM isolation backend exists in the inspected runtime; local OS sandboxing and Docker are separate options.

Verification: sourced
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Happy Agent v2026.11.4 (db813011c038b2288fbc1efd0e10ef2149c838d2), Desktop v0.0.103 (a369b7b0a86a344904a2e171a78e2b0daf0b4be8); targeted first-party pass 2026-11-13; no installed-product test

The scoped first-party source pass does not establish checkout, process. No hands-on scenario was run.

Source: https://happy.engineering/workspaces/
Locator: How Happy creates a workspace
Finding: Parallel workspaces use Git worktrees on branches cut from the remote trunk. File sync and setup commands run per workspace; nongit projects use a plain copy.
Retrieved: 2026-11-08

Source: https://happy.engineering/permissions/
Locator: What the OS sandbox enforces
Finding: The documented OS sandbox uses macOS Seatbelt or Linux namespaces and seccomp. Shell writes, network and host sockets are restricted; file tools can still read credential folders, and MCP servers run outside the sandbox. Docker-backed sessions nest that sandbox inside the container.
Retrieved: 2026-11-08

Source: https://github.com/slopus/happy-agent/blob/db813011c038b2288fbc1efd0e10ef2149c838d2/packages/happy-agent-modules/sources/config/impl/userConfigurationTemplate.ts#L261-L272
Locator: Docker configuration lines 261-272; sources/runners and runtime inspection; search terms docker, virtual machine, microvm
Finding: The scoped runtime config and runner implementation provide host process execution and optional Docker image/container execution, with OS sandbox restrictions. No VM or microVM isolation backend is implemented in the inspected release. A user provisioning a host VM is separate.
Retrieved: 2026-11-13

## happy:RUN-03

Finding: Conversations and child sessions persist across window closure and runtime restart; finished children accept follow-ups with their full context. This is a documentation claim, not a crash test.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/agents/
Locator: Agents That Never Die
Finding: Conversations and child sessions persist across window closure and runtime restart; finished children accept follow-ups with their full context. This is a documentation claim, not a crash test.
Retrieved: 2026-11-08

## happy:RUN-04

Finding: New workspaces copy configured sync files and run setup_commands. They start with a cold build cache. A failed setup command logs a warning while checkout and file sync determine readiness.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/workspaces/
Locator: How Happy creates a workspace
Finding: New workspaces copy configured sync files and run setup_commands. They start with a cold build cache. A failed setup command logs a warning while checkout and file sync determine readiness.
Retrieved: 2026-11-08

## happy:RUN-05

Finding: Smart provider pools can rotate same-provider accounts after authentication or quota-exhaustion errors. Provider quotas are polled about every fifteen minutes and are observations rather than billing records.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/models/
Locator: Use several accounts; Provider usage and quotas
Finding: Smart provider pools can rotate same-provider accounts after authentication or quota-exhaustion errors. Provider quotas are polled about every fifteen minutes and are observations rather than billing records.
Retrieved: 2026-11-08

## happy:RUN-06

Finding: Session status includes active tools, background work, subagents, waiting state, tokens and changed files. Deliberate failure and approval consistency after reconnect remain untested.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/agents/
Locator: What Happy Desktop shows
Finding: Session status includes active tools, background work, subagents, waiting state, tokens and changed files. Deliberate failure and approval consistency after reconnect remain untested.
Retrieved: 2026-11-08

## happy:RUN-07

Finding: Provider quotas are polled about every fifteen minutes and shown in the app; they are observations, not billing records.

Verification: sourced
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/models/
Locator: Provider usage and quotas
Finding: Provider quotas are polled about every fifteen minutes and shown in the app; they are observations, not billing records.
Retrieved: 2026-11-08

## happy:TEAM-01

Finding: Team members see the same projects and sessions and can speak to the agent. Changes in speaker include the member name/email; per-member drafts and task ordering stay personal.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/multiplayer/
Locator: Shared work
Finding: Team members see the same projects and sessions and can speak to the agent. Changes in speaker include the member name/email; per-member drafts and task ordering stay personal.
Retrieved: 2026-11-08

## happy:TEAM-02

Finding: Team mode verifies per-user WorkOS tokens. The owner alone renames the installation, toggles global skills or archives others tasks; members otherwise have equal access and can select Full access. A remove-member action is explicitly not built yet.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/multiplayer/
Locator: Owner; Sign-in; Leaving
Finding: Team mode verifies per-user WorkOS tokens. The owner alone renames the installation, toggles global skills or archives others tasks; members otherwise have equal access and can select Full access. A remove-member action is explicitly not built yet.
Retrieved: 2026-11-08

Source: https://github.com/slopus/slopus.github.io/blob/275790de18910e7a3ff82dab749eb0f4c7cdb427/content/desktop/recepies/enterprise-sign-in.mdx#L1-L15
Locator: content/desktop/recepies/enterprise-sign-in.mdx lines 1-15, commit 275790d
Finding: Recipe for a team-mode Happy Agent whose members sign in through the organization's own identity provider (Okta, Entra ID, Google Workspace, Keycloak or internal SSO) instead of Happy's WorkOS accounts; the daemon verifies JWTs from an organization-owned OAuth authorization server and needs an Agent release with enterprise JWT authentication.
Retrieved: 2026-11-10

Source: https://github.com/slopus/happy-agent/blob/db813011c038b2288fbc1efd0e10ef2149c838d2/packages/happy-agent-modules/sources/config/TeamJwtConfig.ts#L47-L140
Locator: packages/happy-agent-modules/sources/config/TeamJwtConfig.ts lines 47-140, tag v2026.11.4 (commit db81301)
Finding: The released Happy Agent v2026.11.4 parses a [feature.team.jwt] machine configuration with an authorization URL and exactly one of a JWKS URL, public key or secret, used to verify organization-issued team tokens.
Retrieved: 2026-11-10

Source: https://github.com/slopus/happy-desktop/blob/a369b7b0a86a344904a2e171a78e2b0daf0b4be8/packages/happy-desktop-electron/sources/main/serverSessions.ts#L318
Locator: packages/happy-desktop-electron/sources/main/serverSessions.ts line 318, tag v0.0.103 (commit a369b7b)
Finding: The released desktop app v0.0.103 queries a Happy Agent server's v0/authentication endpoint to discover its sign-in methods.
Retrieved: 2026-11-10

## happy:TEAM-03

Finding: Read only, Workspace write, Auto and Full access share a permission model. Auto sends boundary-crossing actions to a separate read-only reviewer and grants only that execution if allowed. Reads and workspace edits are not reviewed.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/permissions/
Locator: Four permission modes; Auto review
Finding: Read only, Workspace write, Auto and Full access share a permission model. Auto sends boundary-crossing actions to a separate read-only reviewer and grants only that execution if allowed. Reads and workspace edits are not reviewed.
Retrieved: 2026-11-08

## happy:TEAM-04

Finding: Session transcripts retain provider-specific opaque data; compaction creates a visible message describing replaced content. No immutable audit retention guarantee was verified.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/models/
Locator: Switch models; How compaction works
Finding: Session transcripts retain provider-specific opaque data; compaction creates a visible message describing replaced content. No immutable audit retention guarantee was verified.
Retrieved: 2026-11-08

## happy:TEAM-05

Finding: The desktop interface has conversations, files, diffs and an attachable terminal; current docs describe changed files beside each conversation. Guided batch review was not established.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/changelog/
Locator: Initial core release; Current interface
Finding: The desktop interface has conversations, files, diffs and an attachable terminal; current docs describe changed files beside each conversation. Guided batch review was not established.
Retrieved: 2026-11-08

## happy:TEAM-06

Finding: Named teammates operate on shared sessions and projects on the same team server. This is not a verified transfer preserving every permission and runtime detail after owner disconnect.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/multiplayer/
Locator: Shared work
Finding: Named teammates operate on shared sessions and projects on the same team server. This is not a verified transfer preserving every permission and runtime detail after owner disconnect.
Retrieved: 2026-11-08

## happy:SHIP-01

Finding: Happy Desktop shows files and diffs with conversations, and its documentation includes a recipe for a live, shared HTML artifact. No exact-revision cross-task gallery test was run.

Verification: sourced
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/changelog/
Locator: Initial core release; Current interface
Finding: The desktop interface has conversations, files, diffs and an attachable terminal; current docs describe changed files beside each conversation. Guided batch review was not established.
Retrieved: 2026-11-08

Source: https://happy.engineering/chief-of-staff/
Locator: Artifacts
Finding: The Chief of Staff documentation links a recipe for building a live, shared HTML artifact. No exact-revision cross-task gallery test was run.
Retrieved: 2026-11-08

## happy:SHIP-03

Finding: The pinned runtime and clients manage Git workspaces and file changes. They have no built-in PR creation or management path; users can ask a coding agent to use external shell or MCP tools.

Verification: sourced
Checked: 2026-11-13
Scheduled recheck: 2027-03-13
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Happy Agent v2026.11.4 (db813011c038b2288fbc1efd0e10ef2149c838d2), Desktop v0.0.103 (a369b7b0a86a344904a2e171a78e2b0daf0b4be8); targeted first-party pass 2026-11-13; no installed-product test

The scoped first-party source pass does not establish branches, worktrees, commits, comments, ci_response, rebase, conflicts, product_state. No hands-on scenario was run.

Source: https://github.com/slopus/happy-agent/blob/db813011c038b2288fbc1efd0e10ef2149c838d2/packages/happy-agent-modules/sources/git/GitModule.ts
Locator: GitModule.ts and sibling sources/git, sources/api, sources/tasks; happy-desktop AppHappyAgentView.tsx at a369b7b; happy CLI/iOS/Android sources at df13fc0. Searches: pull request, pull_request, gh pr, pr create.
Finding: The pinned runtime Git module and clients provide file changes, branches and workspaces, but no product-owned pull-request creation or management path was found. PRs can still be operated through an external shell command or MCP service; that is not a built-in Happy PR workflow.
Retrieved: 2026-11-13

## happy:OWN-01

Finding: The mobile app, original CLI and relay repository has the MIT license. Happy Desktop documentation also explicitly describes its app/runtime as MIT. Hosted and provider services remain separate.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2027-02-06
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://github.com/slopus/happy/blob/df13fc0678ec4093b85fe97c0c48ec9422f5a7c2/LICENSE
Locator: MIT License
Finding: The mobile app, original CLI and relay repository has the MIT license. Happy Desktop documentation also explicitly describes its app/runtime as MIT. Hosted and provider services remain separate.
Retrieved: 2026-11-08

## happy:OWN-02

Finding: The core Happy Agent runs self-hosted on user machines or servers, with Desktop connecting remotely. Happy Social/WorkOS team sign-in and the mobile relay are distinct service dependencies.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/multiplayer/
Locator: How a Happy team server works
Finding: The core Happy Agent runs self-hosted on user machines or servers, with Desktop connecting remotely. Happy Social/WorkOS team sign-in and the mobile relay are distinct service dependencies.
Retrieved: 2026-11-08

## happy:OWN-03

Finding: Happy apps send pseudonymous usage events to PostHog, excluding message content, prompts, code and file names, and offer a Settings opt-out that stops them; the desktop app has its own switch. Happy Agent itself sends no analytics.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-12-08
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://github.com/slopus/happy/blob/df13fc0678ec4093b85fe97c0c48ec9422f5a7c2/PRIVACY.md
Locator: Analytics: PostHog
Finding: Happy apps send pseudonymous product usage events to PostHog, excluding message content, prompts, code and file names, and offer a Settings opt-out that stops them; the desktop app has its own switch. Happy Agent itself sends no analytics.
Retrieved: 2026-11-08

## happy:OWN-04

Finding: Happy is free and MIT licensed. Model use is paid through existing subscriptions and user compute. Phone voice has twenty free minutes per thirty days and an optional paid allowance; custom ElevenLabs use is separate.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-11-22
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/welcome/
Locator: What does Happy cost?
Finding: Happy is free and MIT licensed. Model use is paid through existing subscriptions and user compute. Phone voice has twenty free minutes per thirty days and an optional paid allowance; custom ElevenLabs use is separate.
Retrieved: 2026-11-08

## happy:OWN-06

Finding: Happy Agent v0.4.108 adds GPT-6.1 Terra and stop-state persistence; Desktop v0.0.102 has opt-in beta Stem cells. Current release activity does not imply all experimental UI is enabled by default.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-11-22
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/changelog/
Locator: October 26, 2026
Finding: Happy Agent v0.4.108 adds GPT-6.1 Terra and stop-state persistence; Desktop v0.0.102 has opt-in beta Stem cells. Current release activity does not imply all experimental UI is enabled by default.
Retrieved: 2026-11-08

## happy:OWN-08

Finding: The desktop application and self-hosted runtime are free. Model subscriptions and host compute are separate costs; optional phone voice has its own allowance.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-11-22
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/welcome/
Locator: What does Happy cost?
Finding: The desktop application and self-hosted runtime are free. Model subscriptions and host compute are separate costs; optional phone voice has its own allowance.
Retrieved: 2026-11-08

## happy:OWN-09

Finding: The core ADE has no software subscription charge; coding-model subscriptions, host resources and optional voice are separately scoped. No standardized task-scenario total was invented.

Verification: agent-cross-checked
Checked: 2026-11-08
Scheduled recheck: 2026-11-22
Plan: Not specified
Surface: Happy Desktop with Happy Agent runtime; native mobile apps, original CLI and self-hosted team mode where named
Version: Public documentation retrieved 2026-11-08

Documentation evidence only. Remaining atomic questions and enforcement under failure have not been established by a bounded workflow test.

Source: https://happy.engineering/welcome/
Locator: What does Happy cost?
Finding: The core ADE has no software subscription charge; coding-model subscriptions, host resources and optional voice are separately scoped. No standardized task-scenario total was invented.
Retrieved: 2026-11-08