ralph-claude-code
Open-source Mac and Linux app. Agents run on your computer.
Product websitegithub.com Compare with others Missing or wrong data?
- Desktop app
- Mac and Linux
- Browser
- No
- Phone
- No
- Command line
- Yes
- Agents
- Claude Code
- Your subscription
- Works with your own plan
- Agents run
- On your computer
- Laptop closed
- Not listed
- Isolation
- Container and VM
- Parallel agents
- No
- Scheduled runs
- No
- Coordinator agent
- No
- Built-in issues
- Yes
- Pull requests
- Yes
- Review screen
- No
- Team
- No
- Self-hosting
- Yes
- Licence
- MIT
- Price from
- Free
- How it charges
- No paid plan and bring your own key
- Paid plans from
- Not listed
Receipts
- Checked
- · oldest answer 2026-11-08
- Findings
- 38 published · 29 cross-checked by a second agent · read them all
- Sources
- 18 public pages on 2 sites: github.com, docs.e2b.dev
- Screenshot
- github.com, retrieved · self captured · shows appearance, not capability
- Method
- Same rules for all 103 researched ADEs. How we check · Report wrong data
What it does
Where you use it
Ralph for Claude Code supplies terminal CLI commands and a tmux status monitor on documented macOS/Linux installations, without a product browser or phone client.
Source and scope for Where you use itAgents it runs
The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress. Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
Source and scope for Agents it runsAccounts and subscriptions
Docker execution accepts ANTHROPIC_API_KEY or a scoped copy of existing Claude CLI credentials.
Source and scope for Accounts and subscriptionsWhere agents run
The loop and default Claude execution run on the current host. --sandbox e2b moves Claude execution to E2B while orchestration and file sync stay on the host. tmux detach keeps Ralph running on that same host; remote E2B execution does not move host orchestration.
Source and scope for Where agents runHow work is kept apart
Ralph offers optional Docker execution and an E2B Linux-VM sandbox. The host loop retains orchestration and file sync; it does not provision Git worktrees.
Source and scope for How work is kept apartParallel work
Ralph queue processing is intentionally serial, with one item at a time on a single branch.
Source and scope for Parallel workScheduled runs
Ralph checks completion inside a manually started loop; the product has no schedule or external-event intake that starts new work.
Source and scope for Scheduled runsGit and pull requests
Queue processing commits one commit per issue on the current branch. With --github-issue and --create-pr, Ralph opens a pull request through the gh CLI on completion, optionally as a draft linked to the issue, and can comment on and close the issue.
Source and scope for Git and pull requestsLicence and source
The distributed shell application source is MIT licensed.
Source and scope for Licence and sourcePrice
ralph-claude-code is MIT open source with no paid plan; it uses your own Claude credentials or Anthropic API key, and optional E2B sandboxes are billed by E2B (the tool only estimates and caps that cost).
Source and scope for PriceTyped filters
Every value links to its receipt below. Read plan and surface restrictions before using a filter to decide.
| Filter | Value |
|---|---|
| Desktop OS: macOS | Yes |
| Desktop OS: Windows | Unknown |
| Desktop OS: Linux | Yes |
| Web client | No |
| Phone access: Native iOS | No |
| Phone access: Native Android | No |
| Phone access: Phone browser | No |
| CLI | Yes cross-checked |
| Coding agents: Claude Code | Yes cross-checked |
| Coding agents: Codex | No |
| Coding agents: Cursor Agent | No |
| Coding agents: Gemini CLI | No |
| Coding agents: OpenCode | No |
| Coding agents: Pi | No |
| Coding agents: Grok | No |
| Coding agents: Aider | No |
| Coding agents: Antigravity | No |
| Use my subscription | Yes cross-checked |
| Where agents run: On your computer | Yes cross-checked |
| Where agents run: On your own server | Unknown |
| Where agents run: In the vendor's cloud | No |
| Workspace isolation: Git worktree | No |
| Workspace isolation: Container | Yes |
| Workspace isolation: VM or microVM | Yes |
| Parallel runs | No |
| Coordinator | No |
| Native issue tracking | Yes cross-checked |
| Unattended triggers | No |
| Shared prompting | No |
| Team roles | No |
| SSO | No |
| Pull request workflow | Yes cross-checked |
| Review interface | No |
| Product license: Apache-2.0 | No. MIT cross-checked |
| Product license: MIT | Yes. MIT cross-checked |
| Product license: GPL-3.0 | No. MIT cross-checked |
| Product license: GPL-3.0-or-later | No. MIT cross-checked |
| Product license: AGPL-3.0 | No. MIT cross-checked |
| Product license: AGPL-3.0-or-later | No. MIT cross-checked |
| Product license: Elastic License 2.0 | No. MIT cross-checked |
| Product license: Custom terms | No. MIT cross-checked |
| Product license: Proprietary | No. MIT cross-checked |
| Self-hosting | Yes cross-checked |
| Telemetry opt-out | No |
| Free access | Yes cross-checked |
| Starting price: Free | Yes. Free cross-checked |
| Starting price: Up to $20/month | No. Free cross-checked |
| Starting price: $21–40/month | No. Free cross-checked |
| Starting price: $41–100/month | No. Free cross-checked |
| Starting price: Over $100/month | No. Free cross-checked |
| Starting price: Contact for pricing | No. Free cross-checked |
| Pricing model: Free, no paid plan | Yes cross-checked |
| Pricing model: Free plan | No |
| Pricing model: Subscription | No |
| Pricing model: Per seat | No |
| Pricing model: Usage-based | No |
| Pricing model: Per machine or compute | No |
| Pricing model: Bring your own key | Yes cross-checked |
| Pricing model: One-time purchase | No |
| Paid plans from: No paid plan | Unknown |
| Paid plans from: $10 or less/month | Unknown |
| Paid plans from: $11–20/month | Unknown |
| Paid plans from: $21–40/month | Unknown |
| Paid plans from: $41–100/month | Unknown |
| Paid plans from: Over $100/month | Unknown |
| Paid plans from: Contact sales | Unknown |
| Lifecycle status: Active | Unknown |
| Lifecycle status: Sunsetting | Unknown |
| Keeps going with laptop closed | Unknown |
| Open-source license | Yes. MIT cross-checked |
| Any desktop app | Yes |
| Paid plan up to $20/month | Unknown |
| Paid plan up to $40/month | Unknown |
| Any phone client | No |
Filter receipts
Desktop OS: macOS: Yes
Desktop OS: macOS
Typed value: Yes.
Ralph for Claude Code supplies terminal CLI commands and a tmux status monitor on documented macOS/Linux installations, without a product browser or phone client.
- Cli: The installer provides ralph, ralph-monitor, ralph-queue, ralph-import and project setup commands.
- Notifications: --notify enables macOS/Linux desktop notifications or terminal bell for loop events.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: desktop_os, editor_extensions, task_actions, cross_device_continuity.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
Desktop OS: Linux: Yes
Desktop OS: Linux
Typed value: Yes.
Ralph for Claude Code supplies terminal CLI commands and a tmux status monitor on documented macOS/Linux installations, without a product browser or phone client.
- Cli: The installer provides ralph, ralph-monitor, ralph-queue, ralph-import and project setup commands.
- Notifications: --notify enables macOS/Linux desktop notifications or terminal bell for loop events.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: desktop_os, editor_extensions, task_actions, cross_device_continuity.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
Web client: No
Web client
Typed value: No.
Ralph for Claude Code supplies terminal CLI commands and a tmux status monitor on documented macOS/Linux installations, without a product browser or phone client.
- Mobile apps: The inspected product is a Bash CLI and terminal monitor without native iOS or Android apps.
- Browser: The inspected product is a Bash CLI and terminal monitor without a browser work interface.
- Mobile browser: The inspected product is a Bash CLI and terminal monitor without a phone-browser client.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: desktop_os, editor_extensions, task_actions, cross_device_continuity.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
Phone access: Native iOS: No
Phone access: Native iOS
Typed value: No.
Ralph for Claude Code supplies terminal CLI commands and a tmux status monitor on documented macOS/Linux installations, without a product browser or phone client.
- Mobile apps: The inspected product is a Bash CLI and terminal monitor without native iOS or Android apps.
- Browser: The inspected product is a Bash CLI and terminal monitor without a browser work interface.
- Mobile browser: The inspected product is a Bash CLI and terminal monitor without a phone-browser client.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: desktop_os, editor_extensions, task_actions, cross_device_continuity.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
Phone access: Native Android: No
Phone access: Native Android
Typed value: No.
Ralph for Claude Code supplies terminal CLI commands and a tmux status monitor on documented macOS/Linux installations, without a product browser or phone client.
- Mobile apps: The inspected product is a Bash CLI and terminal monitor without native iOS or Android apps.
- Browser: The inspected product is a Bash CLI and terminal monitor without a browser work interface.
- Mobile browser: The inspected product is a Bash CLI and terminal monitor without a phone-browser client.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: desktop_os, editor_extensions, task_actions, cross_device_continuity.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
Phone access: Phone browser: No
Phone access: Phone browser
Typed value: No.
Ralph for Claude Code supplies terminal CLI commands and a tmux status monitor on documented macOS/Linux installations, without a product browser or phone client.
- Mobile apps: The inspected product is a Bash CLI and terminal monitor without native iOS or Android apps.
- Browser: The inspected product is a Bash CLI and terminal monitor without a browser work interface.
- Mobile browser: The inspected product is a Bash CLI and terminal monitor without a phone-browser client.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: desktop_os, editor_extensions, task_actions, cross_device_continuity.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
CLI: Yes cross-checked
CLI
Typed value: Yes.
The product CLI inspects state, manages queues, imports work and controls loop settings. ralph-queue add/status/next/reorder/validate/remove/clear/process manage persistent queued work.
- Cli: The product CLI inspects state, manages queues, imports work and controls loop settings.
- Operations: ralph-queue add/status/next/reorder/validate/remove/clear/process manage persistent queued work.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code docs/CLI_OPTIONS.md
Finding: The CLI configures hourly calls, timeouts, tool allowlists, session resets and queue processing. Repository-controlled ALLOWED_TOOLS is explicitly not a security boundary.
-
Ralph for Claude Code docs/QUEUE_MANAGEMENT.md
Finding: A persistent queue imports GitHub issues or local PRDs, tracks states and priorities, enforces in-queue dependencies, skips duplicate IDs and processes items sequentially on the current branch.
Coding agents: Claude Code: Yes cross-checked
Coding agents: Claude Code
Typed value: Yes.
The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress. Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Harnesses: The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress.
- Session integration: Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
Coding agents: Codex: No
Coding agents: Codex
Typed value: No.
The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress. Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Harnesses: The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress.
- Session integration: Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
-
Ralph for Claude Code docs/adr/0001-multi-provider-agent-abstraction.md
Finding: Multi-provider abstraction is a planning item. It does not establish released Codex, Gemini or OpenCode support.
Coding agents: Cursor Agent: No
Coding agents: Cursor Agent
Typed value: No.
The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress. Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Harnesses: The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress.
- Session integration: Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
-
Ralph for Claude Code docs/adr/0001-multi-provider-agent-abstraction.md
Finding: Multi-provider abstraction is a planning item. It does not establish released Codex, Gemini or OpenCode support.
Coding agents: Gemini CLI: No
Coding agents: Gemini CLI
Typed value: No.
The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress. Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Harnesses: The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress.
- Session integration: Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
-
Ralph for Claude Code docs/adr/0001-multi-provider-agent-abstraction.md
Finding: Multi-provider abstraction is a planning item. It does not establish released Codex, Gemini or OpenCode support.
Coding agents: OpenCode: No
Coding agents: OpenCode
Typed value: No.
The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress. Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Harnesses: The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress.
- Session integration: Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
-
Ralph for Claude Code docs/adr/0001-multi-provider-agent-abstraction.md
Finding: Multi-provider abstraction is a planning item. It does not establish released Codex, Gemini or OpenCode support.
Coding agents: Pi: No
Coding agents: Pi
Typed value: No.
The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress. Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Harnesses: The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress.
- Session integration: Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
-
Ralph for Claude Code docs/adr/0001-multi-provider-agent-abstraction.md
Finding: Multi-provider abstraction is a planning item. It does not establish released Codex, Gemini or OpenCode support.
Coding agents: Grok: No
Coding agents: Grok
Typed value: No.
The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress. Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Harnesses: The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress.
- Session integration: Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
-
Ralph for Claude Code docs/adr/0001-multi-provider-agent-abstraction.md
Finding: Multi-provider abstraction is a planning item. It does not establish released Codex, Gemini or OpenCode support.
Coding agents: Aider: No
Coding agents: Aider
Typed value: No.
The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress. Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Harnesses: The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress.
- Session integration: Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
-
Ralph for Claude Code docs/adr/0001-multi-provider-agent-abstraction.md
Finding: Multi-provider abstraction is a planning item. It does not establish released Codex, Gemini or OpenCode support.
Coding agents: Antigravity: No
Coding agents: Antigravity
Typed value: No.
The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress. Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Harnesses: The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress.
- Session integration: Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
-
Ralph for Claude Code docs/adr/0001-multi-provider-agent-abstraction.md
Finding: Multi-provider abstraction is a planning item. It does not establish released Codex, Gemini or OpenCode support.
Use my subscription: Yes cross-checked
Use my subscription
Typed value: Yes.
Docker execution accepts ANTHROPIC_API_KEY or a scoped copy of existing Claude CLI credentials.
- Agent subscriptions: Ralph runs the installed claude CLI with its own sign-in on the host; the Docker sandbox copies the user's Claude CLI credentials file when no API key is set. Subscription plan terms were not checked; not tested.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph README: how credentials reach the container
Finding: On the host Ralph invokes the installed claude CLI (default command claude), which uses its own sign-in. In the Docker sandbox it passes ANTHROPIC_API_KEY when set, otherwise copies the user's ~/.claude/.credentials.json into a container-scoped home.
Where agents run: On your computer: Yes cross-checked
Where agents run: On your computer
Typed value: Yes.
The loop and default Claude execution run on the current host. --sandbox e2b moves Claude execution to E2B while orchestration and file sync stay on the host. tmux detach keeps Ralph running on that same host; remote E2B execution does not move host orchestration.
- Local: The loop and default Claude execution run on the current host.
- Client disconnect: tmux detach keeps Ralph running on that same host; remote E2B execution does not move host orchestration.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
Where agents run: In the vendor's cloud: No
Where agents run: In the vendor's cloud
Typed value: No.
The loop and default Claude execution run on the current host. --sandbox e2b moves Claude execution to E2B while orchestration and file sync stay on the host. tmux detach keeps Ralph running on that same host; remote E2B execution does not move host orchestration.
- Local: The loop and default Claude execution run on the current host.
- Managed cloud: --sandbox e2b moves Claude execution to E2B while orchestration and file sync stay on the host.
- Client disconnect: tmux detach keeps Ralph running on that same host; remote E2B execution does not move host orchestration.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
-
Ralph for Claude Code docs/E2B_SANDBOX.md
Finding: E2B runs Claude remotely while the host loop controls iterations and file sync. Sandbox expiry can recreate/reupload; cost limits use estimates rather than invoices.
Workspace isolation: Git worktree: No
Workspace isolation: Git worktree
Typed value: No.
Ralph offers optional Docker execution and an E2B Linux-VM sandbox. The host loop retains orchestration and file sync; it does not provision Git worktrees.
- Worktree: The loop uses the existing checkout and optional backup branches; there is no product-managed Git worktree creation.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: checkout, branch, os_sandbox, filesystem, process, secrets, code_separation, security_boundary.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
Workspace isolation: Container: Yes
Workspace isolation: Container
Typed value: Yes.
Ralph offers optional Docker execution and an E2B Linux-VM sandbox. The host loop retains orchestration and file sync; it does not provision Git worktrees.
- Container: --sandbox docker uses a persistent container with the project bind-mounted read-write.
- Network: Docker supports bridge, none or explicit host networking; repository config is restricted to safer stock choices.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: checkout, branch, os_sandbox, filesystem, process, secrets, code_separation, security_boundary.
-
Ralph for Claude Code docs/DOCKER_SANDBOX.md
Finding: Docker executes Claude inside a persistent container with a read-write project bind mount. Host orchestration remains local, CPU/memory/network are configurable, and credentials are scoped copies or API keys.
Workspace isolation: VM or microVM: Yes
Workspace isolation: VM or microVM
Typed value: Yes.
Ralph offers optional Docker execution and an E2B Linux-VM sandbox. The host loop retains orchestration and file sync; it does not provision Git worktrees.
- Vm: The optional E2B adapter executes Claude in the E2B Sandbox object, which E2B documents as an isolated Linux VM.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: checkout, branch, os_sandbox, filesystem, process, secrets, code_separation, security_boundary.
-
Ralph for Claude Code docs/E2B_SANDBOX.md
Finding: E2B runs Claude remotely while the host loop controls iterations and file sync. Sandbox expiry can recreate/reupload; cost limits use estimates rather than invoices.
-
ralph-claude-code: E2B building blocks, Sandbox
Finding: E2B documents its Sandbox object as an on-demand isolated Linux VM. Ralph optional --sandbox e2b invokes this Sandbox execution path; Ralph orchestration and file sync remain on the host.
Parallel runs: No
Parallel runs
Typed value: No.
Ralph queue processing is intentionally serial, with one item at a time on a single branch.
- Concurrent sessions: The first-party queue documentation explicitly excludes concurrent/parallel processing.
- Concurrent tasks: Queue items are processed one at a time.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: quotas, observed_concurrency, status_visibility.
-
Finding: The queue documentation explicitly excludes concurrent parallel processing and processes items one at a time on a single branch. The loop invokes Claude in sequence rather than running an independent fleet of sessions.
Coordinator: No
Coordinator
Typed value: No.
Ralph repeats one Claude Code loop without a product-supplied agent coordinator or independent agent-session delegation/message tools.
- Orchestration owner: The inspected product supplies a sequential host loop, not a coordinator-agent role with independent-worker delegation tools.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: roles, model_harness_effort, messages, results_return, cross_harness, activation, authentication, live_progress, retained_results, followup, invocation.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
Native issue tracking: Yes cross-checked
Native issue tracking
Typed value: Yes.
Local PRD queue entries can exist before explicit processing starts. Queue priorities come from labels and support persisted reorder. Adding an item is separate from --process-queue or --resume-queue. Queue items retain pending, processing, completed, failed and skipped states.
- Create: Local PRD queue entries can exist before explicit processing starts.
- Prioritize: Queue priorities come from labels and support persisted reorder.
- Explicit dispatch: Adding an item is separate from --process-queue or --resume-queue.
- Durable states: Queue items retain pending, processing, completed, failed and skipped states.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code docs/QUEUE_MANAGEMENT.md
Finding: A persistent queue imports GitHub issues or local PRDs, tracks states and priorities, enforces in-queue dependencies, skips duplicate IDs and processes items sequentially on the current branch.
Unattended triggers: No
Unattended triggers
Typed value: No.
Ralph checks completion inside a manually started loop; the product has no schedule or external-event intake that starts new work.
- Schedules: No schedule runner is implemented in the inspected CLI/runtime.
- Events: Notifications are outgoing; there is no incoming event trigger for agent starts.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: recurring_tasks, scheduler_location, overlaps, retries, client_independence.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
Shared prompting: No
Shared prompting
Typed value: No.
The product is a local CLI/terminal monitor without a shared multi-human prompting session.
- Coprompting: The inspected CLI/runtime exposes no multi-human prompting service or shared session input model.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: visibility, discussion, comments, presence, ordering, conflicts, attribution, permissions, plan_restrictions, owner_online.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
Team roles: No
Team roles
Typed value: No.
Ralph has no product-owned human identity service, named team roles or enterprise SSO.
- Organization roles: No human organization/workspace role-permission model exists in the inspected CLI.
- Sso: The product has no enterprise SSO implementation; Claude/provider logins remain separate.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: project_roles, task_roles, membership, provisioning, unauthorized_access, revocation, plan_scope.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
SSO: No
SSO
Typed value: No.
Ralph has no product-owned human identity service, named team roles or enterprise SSO.
- Organization roles: No human organization/workspace role-permission model exists in the inspected CLI.
- Sso: The product has no enterprise SSO implementation; Claude/provider logins remain separate.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: project_roles, task_roles, membership, provisioning, unauthorized_access, revocation, plan_scope.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
Pull request workflow: Yes cross-checked
Pull request workflow
Typed value: Yes.
Queue processing commits one commit per issue on the current branch. With --github-issue and --create-pr, Ralph opens a pull request through the gh CLI on completion, optionally as a draft linked to the issue, and can comment on and close the issue.
- Pull requests: --create-pr opens a pull request through the gh CLI when the loop completes; --draft-pr and --link-issue set draft state and a Closes #N link.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph README: GitHub issue lifecycle
Finding: With --github-issue, Ralph uses the gh CLI to post progress comments, create a pull request on completion (optionally as a draft and linked with Closes #N), post a summary, close the issue and open follow-up issues.
Review interface: No
Review interface
Typed value: No.
Ralph monitor displays loop status and logs without proposed code changes or a code-diff review UI.
- Change understanding: The monitor is display-only loop status/log output, with no code-diff review interface.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: pending_reviews, evidence_navigation, comments, request_changes, shipping_guidance, effort, missed_evidence, keyboard, narrow_screen, context_switching, roles, retained_context.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
Product license: Apache-2.0: No. MIT cross-checked
Product license: Apache-2.0
Typed value: No. MIT.
The distributed shell application source is MIT licensed.
- Client license: The distributed shell application source is MIT licensed.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Finding: Ralph repository code is MIT licensed.
Product license: MIT: Yes. MIT cross-checked
Product license: MIT
Typed value: Yes. MIT.
The distributed shell application source is MIT licensed.
- Client license: The distributed shell application source is MIT licensed.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Finding: Ralph repository code is MIT licensed.
Product license: GPL-3.0: No. MIT cross-checked
Product license: GPL-3.0
Typed value: No. MIT.
The distributed shell application source is MIT licensed.
- Client license: The distributed shell application source is MIT licensed.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Finding: Ralph repository code is MIT licensed.
Product license: GPL-3.0-or-later: No. MIT cross-checked
Product license: GPL-3.0-or-later
Typed value: No. MIT.
The distributed shell application source is MIT licensed.
- Client license: The distributed shell application source is MIT licensed.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Finding: Ralph repository code is MIT licensed.
Product license: AGPL-3.0: No. MIT cross-checked
Product license: AGPL-3.0
Typed value: No. MIT.
The distributed shell application source is MIT licensed.
- Client license: The distributed shell application source is MIT licensed.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Finding: Ralph repository code is MIT licensed.
Product license: AGPL-3.0-or-later: No. MIT cross-checked
Product license: AGPL-3.0-or-later
Typed value: No. MIT.
The distributed shell application source is MIT licensed.
- Client license: The distributed shell application source is MIT licensed.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Finding: Ralph repository code is MIT licensed.
Product license: Elastic License 2.0: No. MIT cross-checked
Product license: Elastic License 2.0
Typed value: No. MIT.
The distributed shell application source is MIT licensed.
- Client license: The distributed shell application source is MIT licensed.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Finding: Ralph repository code is MIT licensed.
Product license: Custom terms: No. MIT cross-checked
Product license: Custom terms
Typed value: No. MIT.
The distributed shell application source is MIT licensed.
- Client license: The distributed shell application source is MIT licensed.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Finding: Ralph repository code is MIT licensed.
Product license: Proprietary: No. MIT cross-checked
Product license: Proprietary
Typed value: No. MIT.
The distributed shell application source is MIT licensed.
- Client license: The distributed shell application source is MIT licensed.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Finding: Ralph repository code is MIT licensed.
Self-hosting: Yes cross-checked
Self-hosting
Typed value: Yes.
The loop, queue and monitor run locally from the public source distribution. Claude authentication is required; optional E2B adds separate cloud credentials and compute costs.
- Self hosting: The loop, queue and monitor run locally from the public source distribution.
- Vendor dependencies: Claude authentication is required; optional E2B adds separate cloud credentials and compute costs.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
Telemetry opt-out: No
Telemetry opt-out
Typed value: No.
The inspected Ralph CLI implements no ADE analytics reporter or telemetry opt-out switch. Selected-agent, GitHub, sandbox and user-configured notification traffic remain separate.
- Telemetry disclosure: No product analytics reporter is implemented in the inspected runtime/CLI.
- Telemetry controls: There is no Ralph telemetry opt-out switch because the inspected product has no ADE telemetry subsystem.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: code_data, prompts, credentials, logs, artifacts, retention, deletion, training, residency, subprocessors, telemetry_defaults, consent_prompt, provider_separation.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
Free access: Yes cross-checked
Free access
Typed value: Yes.
Ralph is a free MIT-licensed shell tool installed from the public repository; no paid plan is shown. Claude usage and optional E2B sandboxes are billed by their providers.
- Ongoing free: The tool is a free MIT-licensed install from the public repository.
- Paid only: Not paid-only: the tool installs free of charge from the public repository.
- Underlying Claude plan limits were not checked. No installed-product test was run.
-
Finding: Ralph is installed globally from a clone of the public repository with ./install.sh; the README names no price, account or paid plan.
-
Finding: The repository is under the MIT License, copyright Frank Bria.
Starting price: Free: Yes. Free cross-checked
Starting price: Free
Typed value: Yes. Free.
Ralph is a free MIT-licensed shell tool installed from the public repository; no paid plan is shown. Claude usage and optional E2B sandboxes are billed by their providers.
- Ongoing free: The tool is a free MIT-licensed install from the public repository.
- Paid only: Not paid-only: the tool installs free of charge from the public repository.
- Underlying Claude plan limits were not checked. No installed-product test was run.
-
Finding: Ralph is installed globally from a clone of the public repository with ./install.sh; the README names no price, account or paid plan.
-
Finding: The repository is under the MIT License, copyright Frank Bria.
Starting price: Up to $20/month: No. Free cross-checked
Starting price: Up to $20/month
Typed value: No. Free.
Ralph is a free MIT-licensed shell tool installed from the public repository; no paid plan is shown. Claude usage and optional E2B sandboxes are billed by their providers.
- Ongoing free: The tool is a free MIT-licensed install from the public repository.
- Paid only: Not paid-only: the tool installs free of charge from the public repository.
- Underlying Claude plan limits were not checked. No installed-product test was run.
-
Finding: Ralph is installed globally from a clone of the public repository with ./install.sh; the README names no price, account or paid plan.
-
Finding: The repository is under the MIT License, copyright Frank Bria.
Starting price: $21–40/month: No. Free cross-checked
Starting price: $21–40/month
Typed value: No. Free.
Ralph is a free MIT-licensed shell tool installed from the public repository; no paid plan is shown. Claude usage and optional E2B sandboxes are billed by their providers.
- Ongoing free: The tool is a free MIT-licensed install from the public repository.
- Paid only: Not paid-only: the tool installs free of charge from the public repository.
- Underlying Claude plan limits were not checked. No installed-product test was run.
-
Finding: Ralph is installed globally from a clone of the public repository with ./install.sh; the README names no price, account or paid plan.
-
Finding: The repository is under the MIT License, copyright Frank Bria.
Starting price: $41–100/month: No. Free cross-checked
Starting price: $41–100/month
Typed value: No. Free.
Ralph is a free MIT-licensed shell tool installed from the public repository; no paid plan is shown. Claude usage and optional E2B sandboxes are billed by their providers.
- Ongoing free: The tool is a free MIT-licensed install from the public repository.
- Paid only: Not paid-only: the tool installs free of charge from the public repository.
- Underlying Claude plan limits were not checked. No installed-product test was run.
-
Finding: Ralph is installed globally from a clone of the public repository with ./install.sh; the README names no price, account or paid plan.
-
Finding: The repository is under the MIT License, copyright Frank Bria.
Starting price: Over $100/month: No. Free cross-checked
Starting price: Over $100/month
Typed value: No. Free.
Ralph is a free MIT-licensed shell tool installed from the public repository; no paid plan is shown. Claude usage and optional E2B sandboxes are billed by their providers.
- Ongoing free: The tool is a free MIT-licensed install from the public repository.
- Paid only: Not paid-only: the tool installs free of charge from the public repository.
- Underlying Claude plan limits were not checked. No installed-product test was run.
-
Finding: Ralph is installed globally from a clone of the public repository with ./install.sh; the README names no price, account or paid plan.
-
Finding: The repository is under the MIT License, copyright Frank Bria.
Starting price: Contact for pricing: No. Free cross-checked
Starting price: Contact for pricing
Typed value: No. Free.
Ralph is a free MIT-licensed shell tool installed from the public repository; no paid plan is shown. Claude usage and optional E2B sandboxes are billed by their providers.
- Ongoing free: The tool is a free MIT-licensed install from the public repository.
- Paid only: Not paid-only: the tool installs free of charge from the public repository.
- Underlying Claude plan limits were not checked. No installed-product test was run.
-
Finding: Ralph is installed globally from a clone of the public repository with ./install.sh; the README names no price, account or paid plan.
-
Finding: The repository is under the MIT License, copyright Frank Bria.
Pricing model: Free, no paid plan: Yes cross-checked
Pricing model: Free, no paid plan
Typed value: Yes.
ralph-claude-code is MIT open source with no paid plan; it uses your own Claude credentials or Anthropic API key, and optional E2B sandboxes are billed by E2B (the tool only estimates and caps that cost).
- First-party pricing pages, terms and licences read 2026-11-10; no purchase, billing or authenticated account test.
-
frankbria/ralph-claude-code LICENSE (MIT)
Finding: The MIT License lets anyone obtain, use, copy and modify the software free of charge.
Pricing model: Free plan: No
Pricing model: Free plan
Typed value: No.
ralph-claude-code is MIT open source with no paid plan; it uses your own Claude credentials or Anthropic API key, and optional E2B sandboxes are billed by E2B (the tool only estimates and caps that cost).
- Model cost: Claude runs on your own ANTHROPIC_API_KEY or Claude login credentials, so Anthropic bills for usage.
- Compute cost: The optional E2B sandbox has its own cost, which the tool estimates from runtime times a user-set hourly rate and can cap. The README gives no E2B prices.
- First-party pricing pages, terms and licences read 2026-11-10; no purchase, billing or authenticated account test.
-
Finding: Says the tool is MIT licensed. Claude auth can use your own ANTHROPIC_API_KEY or Claude credentials. The optional E2B cloud sandbox has cost estimated from runtime times a user-set hourly rate, with budget caps, and E2B charges for it.
-
frankbria/ralph-claude-code LICENSE (MIT)
Finding: The MIT License lets anyone obtain, use, copy and modify the software free of charge.
Pricing model: Subscription: No
Pricing model: Subscription
Typed value: No.
ralph-claude-code is MIT open source with no paid plan; it uses your own Claude credentials or Anthropic API key, and optional E2B sandboxes are billed by E2B (the tool only estimates and caps that cost).
- Model cost: Claude runs on your own ANTHROPIC_API_KEY or Claude login credentials, so Anthropic bills for usage.
- Compute cost: The optional E2B sandbox has its own cost, which the tool estimates from runtime times a user-set hourly rate and can cap. The README gives no E2B prices.
- First-party pricing pages, terms and licences read 2026-11-10; no purchase, billing or authenticated account test.
-
Finding: Says the tool is MIT licensed. Claude auth can use your own ANTHROPIC_API_KEY or Claude credentials. The optional E2B cloud sandbox has cost estimated from runtime times a user-set hourly rate, with budget caps, and E2B charges for it.
-
frankbria/ralph-claude-code LICENSE (MIT)
Finding: The MIT License lets anyone obtain, use, copy and modify the software free of charge.
Pricing model: Per seat: No
Pricing model: Per seat
Typed value: No.
ralph-claude-code is MIT open source with no paid plan; it uses your own Claude credentials or Anthropic API key, and optional E2B sandboxes are billed by E2B (the tool only estimates and caps that cost).
- Model cost: Claude runs on your own ANTHROPIC_API_KEY or Claude login credentials, so Anthropic bills for usage.
- Compute cost: The optional E2B sandbox has its own cost, which the tool estimates from runtime times a user-set hourly rate and can cap. The README gives no E2B prices.
- First-party pricing pages, terms and licences read 2026-11-10; no purchase, billing or authenticated account test.
-
Finding: Says the tool is MIT licensed. Claude auth can use your own ANTHROPIC_API_KEY or Claude credentials. The optional E2B cloud sandbox has cost estimated from runtime times a user-set hourly rate, with budget caps, and E2B charges for it.
-
frankbria/ralph-claude-code LICENSE (MIT)
Finding: The MIT License lets anyone obtain, use, copy and modify the software free of charge.
Pricing model: Usage-based: No
Pricing model: Usage-based
Typed value: No.
ralph-claude-code is MIT open source with no paid plan; it uses your own Claude credentials or Anthropic API key, and optional E2B sandboxes are billed by E2B (the tool only estimates and caps that cost).
- Model cost: Claude runs on your own ANTHROPIC_API_KEY or Claude login credentials, so Anthropic bills for usage.
- Compute cost: The optional E2B sandbox has its own cost, which the tool estimates from runtime times a user-set hourly rate and can cap. The README gives no E2B prices.
- First-party pricing pages, terms and licences read 2026-11-10; no purchase, billing or authenticated account test.
-
Finding: Says the tool is MIT licensed. Claude auth can use your own ANTHROPIC_API_KEY or Claude credentials. The optional E2B cloud sandbox has cost estimated from runtime times a user-set hourly rate, with budget caps, and E2B charges for it.
-
frankbria/ralph-claude-code LICENSE (MIT)
Finding: The MIT License lets anyone obtain, use, copy and modify the software free of charge.
Pricing model: Per machine or compute: No
Pricing model: Per machine or compute
Typed value: No.
ralph-claude-code is MIT open source with no paid plan; it uses your own Claude credentials or Anthropic API key, and optional E2B sandboxes are billed by E2B (the tool only estimates and caps that cost).
- Model cost: Claude runs on your own ANTHROPIC_API_KEY or Claude login credentials, so Anthropic bills for usage.
- Compute cost: The optional E2B sandbox has its own cost, which the tool estimates from runtime times a user-set hourly rate and can cap. The README gives no E2B prices.
- First-party pricing pages, terms and licences read 2026-11-10; no purchase, billing or authenticated account test.
-
Finding: Says the tool is MIT licensed. Claude auth can use your own ANTHROPIC_API_KEY or Claude credentials. The optional E2B cloud sandbox has cost estimated from runtime times a user-set hourly rate, with budget caps, and E2B charges for it.
-
frankbria/ralph-claude-code LICENSE (MIT)
Finding: The MIT License lets anyone obtain, use, copy and modify the software free of charge.
Pricing model: Bring your own key: Yes cross-checked
Pricing model: Bring your own key
Typed value: Yes.
ralph-claude-code is MIT open source with no paid plan; it uses your own Claude credentials or Anthropic API key, and optional E2B sandboxes are billed by E2B (the tool only estimates and caps that cost).
- Model cost: Claude runs on your own ANTHROPIC_API_KEY or Claude login credentials, so Anthropic bills for usage.
- Compute cost: The optional E2B sandbox has its own cost, which the tool estimates from runtime times a user-set hourly rate and can cap. The README gives no E2B prices.
- First-party pricing pages, terms and licences read 2026-11-10; no purchase, billing or authenticated account test.
-
Finding: Says the tool is MIT licensed. Claude auth can use your own ANTHROPIC_API_KEY or Claude credentials. The optional E2B cloud sandbox has cost estimated from runtime times a user-set hourly rate, with budget caps, and E2B charges for it.
Pricing model: One-time purchase: No
Pricing model: One-time purchase
Typed value: No.
ralph-claude-code is MIT open source with no paid plan; it uses your own Claude credentials or Anthropic API key, and optional E2B sandboxes are billed by E2B (the tool only estimates and caps that cost).
- Model cost: Claude runs on your own ANTHROPIC_API_KEY or Claude login credentials, so Anthropic bills for usage.
- Compute cost: The optional E2B sandbox has its own cost, which the tool estimates from runtime times a user-set hourly rate and can cap. The README gives no E2B prices.
- First-party pricing pages, terms and licences read 2026-11-10; no purchase, billing or authenticated account test.
-
Finding: Says the tool is MIT licensed. Claude auth can use your own ANTHROPIC_API_KEY or Claude credentials. The optional E2B cloud sandbox has cost estimated from runtime times a user-set hourly rate, with budget caps, and E2B charges for it.
-
frankbria/ralph-claude-code LICENSE (MIT)
Finding: The MIT License lets anyone obtain, use, copy and modify the software free of charge.
Open-source license: Yes. MIT cross-checked
Open-source license
Typed value: Yes. MIT.
The distributed shell application source is MIT licensed.
- Client license: The distributed shell application source is MIT licensed.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Finding: Ralph repository code is MIT licensed.
Any desktop app: Yes
Any desktop app
Typed value: Yes.
Ralph for Claude Code supplies terminal CLI commands and a tmux status monitor on documented macOS/Linux installations, without a product browser or phone client.
- Cli: The installer provides ralph, ralph-monitor, ralph-queue, ralph-import and project setup commands.
- Notifications: --notify enables macOS/Linux desktop notifications or terminal bell for loop events.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: desktop_os, editor_extensions, task_actions, cross_device_continuity.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
Any phone client: No
Any phone client
Typed value: No.
Ralph for Claude Code supplies terminal CLI commands and a tmux status monitor on documented macOS/Linux installations, without a product browser or phone client.
- Mobile apps: The inspected product is a Bash CLI and terminal monitor without native iOS or Android apps.
- Browser: The inspected product is a Bash CLI and terminal monitor without a browser work interface.
- Mobile browser: The inspected product is a Bash CLI and terminal monitor without a phone-browser client.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: desktop_os, editor_extensions, task_actions, cross_device_continuity.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
All public findings
Findings may summarize a source. They are not presented as verbatim statements.
ralph-claude-code:ACC-01: Ralph for Claude Code supplies terminal CLI commands and a tmux status monitor on documented macOS/Linux installations, without a product browser or phone client.
ralph-claude-code:ACC-01
Ralph for Claude Code supplies terminal CLI commands and a tmux status monitor on documented macOS/Linux installations, without a product browser or phone client.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: desktop_os, editor_extensions, task_actions, cross_device_continuity.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
ralph-claude-code:ACC-02: The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress. Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity. cross-checked
ralph-claude-code:ACC-02
The shipped loop runs Claude Code; multi-provider abstraction is still listed as in progress. Ralph invokes Claude Code repeatedly and uses explicit --resume session IDs for continuity.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
-
Ralph for Claude Code docs/adr/0001-multi-provider-agent-abstraction.md
Finding: Multi-provider abstraction is a planning item. It does not establish released Codex, Gemini or OpenCode support.
ralph-claude-code:ACC-03: Docker execution accepts ANTHROPIC_API_KEY or a scoped copy of existing Claude CLI credentials. cross-checked
ralph-claude-code:ACC-03
Docker execution accepts ANTHROPIC_API_KEY or a scoped copy of existing Claude CLI credentials.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code docs/DOCKER_SANDBOX.md
Finding: Docker executes Claude inside a persistent container with a read-write project bind mount. Host orchestration remains local, CPU/memory/network are configurable, and credentials are scoped copies or API keys.
-
Ralph README: how credentials reach the container
Finding: On the host Ralph invokes the installed claude CLI (default command claude), which uses its own sign-in. In the Docker sandbox it passes ANTHROPIC_API_KEY when set, otherwise copies the user's ~/.claude/.credentials.json into a container-scoped home.
ralph-claude-code:ACC-04: Install once, enable or import a project, review .ralph files and run ralph --monitor. README lists Bash, Claude Code, jq, Git, GNU timeout and optional tmux prerequisites. cross-checked
ralph-claude-code:ACC-04
Install once, enable or import a project, review .ralph files and run ralph --monitor. README lists Bash, Claude Code, jq, Git, GNU timeout and optional tmux prerequisites.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
ralph-claude-code:ACC-05: The product CLI inspects state, manages queues, imports work and controls loop settings. ralph-queue add/status/next/reorder/validate/remove/clear/process manage persistent queued work. cross-checked
ralph-claude-code:ACC-05
The product CLI inspects state, manages queues, imports work and controls loop settings. ralph-queue add/status/next/reorder/validate/remove/clear/process manage persistent queued work.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code docs/CLI_OPTIONS.md
Finding: The CLI configures hourly calls, timeouts, tool allowlists, session resets and queue processing. Repository-controlled ALLOWED_TOOLS is explicitly not a security boundary.
-
Ralph for Claude Code docs/QUEUE_MANAGEMENT.md
Finding: A persistent queue imports GitHub issues or local PRDs, tracks states and priorities, enforces in-queue dependencies, skips duplicate IDs and processes items sequentially on the current branch.
ralph-claude-code:ACC-06: PRD imports populate .ralph prompt, specs and task checklist files. cross-checked
ralph-claude-code:ACC-06
PRD imports populate .ralph prompt, specs and task checklist files.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
-
Ralph for Claude Code docs/user-guide/02-understanding-ralph-files.md
Finding: User-editable prompt/spec files and prioritized checklists provide the loop context and work plan. They are files rather than a documented revision/approval system.
ralph-claude-code:PLAN-01: A local PRD file can be added to the persistent queue without a GitHub issue. GitHub issues import with labels, priority, milestone and parsed dependencies. GitHub issue bodies are fetched again at processing time; this is not documented as bidirectional issue synchronization. cross-checked
ralph-claude-code:PLAN-01
A local PRD file can be added to the persistent queue without a GitHub issue. GitHub issues import with labels, priority, milestone and parsed dependencies. GitHub issue bodies are fetched again at processing time; this is not documented as bidirectional issue synchronization.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code docs/QUEUE_MANAGEMENT.md
Finding: A persistent queue imports GitHub issues or local PRDs, tracks states and priorities, enforces in-queue dependencies, skips duplicate IDs and processes items sequentially on the current branch.
ralph-claude-code:PLAN-02: fix_plan.md holds prioritized checkbox tasks while specs holds detailed requirements. Users edit Markdown plans and specifications directly. The queue JSON persists across restarts; processing items from interrupted runs reset to pending on resume. cross-checked
ralph-claude-code:PLAN-02
fix_plan.md holds prioritized checkbox tasks while specs holds detailed requirements. Users edit Markdown plans and specifications directly. The queue JSON persists across restarts; processing items from interrupted runs reset to pending on resume.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code docs/QUEUE_MANAGEMENT.md
Finding: A persistent queue imports GitHub issues or local PRDs, tracks states and priorities, enforces in-queue dependencies, skips duplicate IDs and processes items sequentially on the current branch.
-
Ralph for Claude Code docs/user-guide/02-understanding-ralph-files.md
Finding: User-editable prompt/spec files and prioritized checklists provide the loop context and work plan. They are files rather than a documented revision/approval system.
ralph-claude-code:PLAN-03: depends on, blocked by and requires references inside issue bodies become queue dependencies. A queued item becomes ready only after all referenced in-queue dependencies complete; circular dependencies are refused. cross-checked
ralph-claude-code:PLAN-03
depends on, blocked by and requires references inside issue bodies become queue dependencies. A queued item becomes ready only after all referenced in-queue dependencies complete; circular dependencies are refused.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code docs/QUEUE_MANAGEMENT.md
Finding: A persistent queue imports GitHub issues or local PRDs, tracks states and priorities, enforces in-queue dependencies, skips duplicate IDs and processes items sequentially on the current branch.
ralph-claude-code:PLAN-04: PROMPT.md, AGENT.md and specs/ supply project requirements and build guidance. Explicit Claude session IDs preserve loop context until configured expiry or reset. cross-checked
ralph-claude-code:PLAN-04
PROMPT.md, AGENT.md and specs/ supply project requirements and build guidance. Explicit Claude session IDs preserve loop context until configured expiry or reset.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
-
Ralph for Claude Code docs/user-guide/02-understanding-ralph-files.md
Finding: User-editable prompt/spec files and prioritized checklists provide the loop context and work plan. They are files rather than a documented revision/approval system.
ralph-claude-code:PLAN-06: Local PRD queue entries can exist before explicit processing starts. Queue priorities come from labels and support persisted reorder. Adding an item is separate from --process-queue or --resume-queue. Queue items retain pending, processing, completed, failed and skipped states. cross-checked
ralph-claude-code:PLAN-06
Local PRD queue entries can exist before explicit processing starts. Queue priorities come from labels and support persisted reorder. Adding an item is separate from --process-queue or --resume-queue. Queue items retain pending, processing, completed, failed and skipped states.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code docs/QUEUE_MANAGEMENT.md
Finding: A persistent queue imports GitHub issues or local PRDs, tracks states and priorities, enforces in-queue dependencies, skips duplicate IDs and processes items sequentially on the current branch.
ralph-claude-code:PLAN-07: Documented CLI reads persistent queue status and next ready item. Agent access to the same CLI is possible but no installed agent-write fixture was run. cross-checked
ralph-claude-code:PLAN-07
Documented CLI reads persistent queue status and next ready item. Agent access to the same CLI is possible but no installed agent-write fixture was run.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code docs/QUEUE_MANAGEMENT.md
Finding: A persistent queue imports GitHub issues or local PRDs, tracks states and priorities, enforces in-queue dependencies, skips duplicate IDs and processes items sequentially on the current branch.
ralph-claude-code:COORD-01: Ralph queue processing is intentionally serial, with one item at a time on a single branch.
ralph-claude-code:COORD-01
Ralph queue processing is intentionally serial, with one item at a time on a single branch.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: quotas, observed_concurrency, status_visibility.
-
Finding: The queue documentation explicitly excludes concurrent parallel processing and processes items one at a time on a single branch. The loop invokes Claude in sequence rather than running an independent fleet of sessions.
ralph-claude-code:COORD-02: Duplicate queue IDs are skipped; ready items run in priority/FIFO order. cross-checked
ralph-claude-code:COORD-02
Duplicate queue IDs are skipped; ready items run in priority/FIFO order.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code docs/QUEUE_MANAGEMENT.md
Finding: A persistent queue imports GitHub issues or local PRDs, tracks states and priorities, enforces in-queue dependencies, skips duplicate IDs and processes items sequentially on the current branch.
ralph-claude-code:COORD-03: Ralph repeats one Claude Code loop without a product-supplied agent coordinator or independent agent-session delegation/message tools.
ralph-claude-code:COORD-03
Ralph repeats one Claude Code loop without a product-supplied agent coordinator or independent agent-session delegation/message tools.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: roles, model_harness_effort, messages, results_return, cross_harness, activation, authentication, live_progress, retained_results, followup, invocation.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
-
Finding: The queue documentation explicitly excludes concurrent parallel processing and processes items one at a time on a single branch. The loop invokes Claude in sequence rather than running an independent fleet of sessions.
ralph-claude-code:COORD-04: Ralph checks completion inside a manually started loop; the product has no schedule or external-event intake that starts new work.
ralph-claude-code:COORD-04
Ralph checks completion inside a manually started loop; the product has no schedule or external-event intake that starts new work.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: recurring_tasks, scheduler_location, overlaps, retries, client_independence.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
ralph-claude-code:COORD-05: Ctrl+C interrupts the monitor; loop termination and session reset are separately documented. Open reports about SIGTERM waits were not resolved by this pass. --notify reports key loop events. cross-checked
ralph-claude-code:COORD-05
Ctrl+C interrupts the monitor; loop termination and session reset are separately documented. Open reports about SIGTERM waits were not resolved by this pass. --notify reports key loop events.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
ralph-claude-code:RUN-01: The loop and default Claude execution run on the current host. --sandbox e2b moves Claude execution to E2B while orchestration and file sync stay on the host. tmux detach keeps Ralph running on that same host; remote E2B execution does not move host orchestration. cross-checked
ralph-claude-code:RUN-01
The loop and default Claude execution run on the current host. --sandbox e2b moves Claude execution to E2B while orchestration and file sync stay on the host. tmux detach keeps Ralph running on that same host; remote E2B execution does not move host orchestration.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
-
Ralph for Claude Code docs/E2B_SANDBOX.md
Finding: E2B runs Claude remotely while the host loop controls iterations and file sync. Sandbox expiry can recreate/reupload; cost limits use estimates rather than invoices.
ralph-claude-code:RUN-02: Ralph offers optional Docker execution and an E2B Linux-VM sandbox. The host loop retains orchestration and file sync; it does not provision Git worktrees.
ralph-claude-code:RUN-02
Ralph offers optional Docker execution and an E2B Linux-VM sandbox. The host loop retains orchestration and file sync; it does not provision Git worktrees.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: checkout, branch, os_sandbox, filesystem, process, secrets, code_separation, security_boundary.
-
Ralph for Claude Code docs/DOCKER_SANDBOX.md
Finding: Docker executes Claude inside a persistent container with a read-write project bind mount. Host orchestration remains local, CPU/memory/network are configurable, and credentials are scoped copies or API keys.
-
Ralph for Claude Code docs/E2B_SANDBOX.md
Finding: E2B runs Claude remotely while the host loop controls iterations and file sync. Sandbox expiry can recreate/reupload; cost limits use estimates rather than invoices.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
-
ralph-claude-code: E2B building blocks, Sandbox
Finding: E2B documents its Sandbox object as an on-demand isolated Linux VM. Ralph optional --sandbox e2b invokes this Sandbox execution path; Ralph orchestration and file sync remain on the host.
ralph-claude-code:RUN-03: E2B expiry can recreate the sandbox and upload project files again. This is distinct from preserving a live model process. Interrupted processing items reset to pending when the queue resumes; completed and failed items remain recorded. Claude sessions expire after 24 hours by default and support explicit reset. cross-checked
ralph-claude-code:RUN-03
E2B expiry can recreate the sandbox and upload project files again. This is distinct from preserving a live model process. Interrupted processing items reset to pending when the queue resumes; completed and failed items remain recorded. Claude sessions expire after 24 hours by default and support explicit reset.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
-
Ralph for Claude Code docs/QUEUE_MANAGEMENT.md
Finding: A persistent queue imports GitHub issues or local PRDs, tracks states and priorities, enforces in-queue dependencies, skips duplicate IDs and processes items sequentially on the current branch.
-
Ralph for Claude Code docs/E2B_SANDBOX.md
Finding: E2B runs Claude remotely while the host loop controls iterations and file sync. Sandbox expiry can recreate/reupload; cost limits use estimates rather than invoices.
ralph-claude-code:RUN-04: E2B bootstraps Claude CLI on the default template or accepts a prebuilt custom template. Docker scopes API-key files or copies native Claude credentials into a temporary per-run directory. cross-checked
ralph-claude-code:RUN-04
E2B bootstraps Claude CLI on the default template or accepts a prebuilt custom template. Docker scopes API-key files or copies native Claude credentials into a temporary per-run directory.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code docs/DOCKER_SANDBOX.md
Finding: Docker executes Claude inside a persistent container with a read-write project bind mount. Host orchestration remains local, CPU/memory/network are configurable, and credentials are scoped copies or API keys.
-
Ralph for Claude Code docs/E2B_SANDBOX.md
Finding: E2B runs Claude remotely while the host loop controls iterations and file sync. Sandbox expiry can recreate/reupload; cost limits use estimates rather than invoices.
ralph-claude-code:RUN-05: Claude invocation timeouts are configurable from 1 to 120 minutes. Circuit-breaker cooldown and auto-reset controls bound repeated no-progress/error loops. Hourly call limits and dual-condition completion checks stop or delay loop work. cross-checked
ralph-claude-code:RUN-05
Claude invocation timeouts are configurable from 1 to 120 minutes. Circuit-breaker cooldown and auto-reset controls bound repeated no-progress/error loops. Hourly call limits and dual-condition completion checks stop or delay loop work.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
ralph-claude-code:RUN-06: The monitor shows local logs, loop count, API-call usage and queue totals from JSON state. Queue monitor reports pending, active, completed and failed item totals. cross-checked
ralph-claude-code:RUN-06
The monitor shows local logs, loop count, API-call usage and queue totals from JSON state. Queue monitor reports pending, active, completed and failed item totals.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code ralph_monitor.sh
Finding: The actual monitor reads local status/progress/queue JSON and log files, displays loop/API-call counts and queue state totals, and refreshes every two seconds.
ralph-claude-code:RUN-07: E2B sandbox cost uses an estimated hourly rate, not a reconciled provider invoice. E2B sandbox estimates, alerts and --sandbox-max-cost are documented separately from model spend. cross-checked
ralph-claude-code:RUN-07
E2B sandbox cost uses an estimated hourly rate, not a reconciled provider invoice. E2B sandbox estimates, alerts and --sandbox-max-cost are documented separately from model spend.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code docs/E2B_SANDBOX.md
Finding: E2B runs Claude remotely while the host loop controls iterations and file sync. Sandbox expiry can recreate/reupload; cost limits use estimates rather than invoices.
ralph-claude-code:RUN-08: The queue is sequential on one branch, not a documented parallel fleet scheduler. Docker defaults to 4g memory and 2 CPUs with explicit override flags. cross-checked
ralph-claude-code:RUN-08
The queue is sequential on one branch, not a documented parallel fleet scheduler. Docker defaults to 4g memory and 2 CPUs with explicit override flags.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code docs/QUEUE_MANAGEMENT.md
Finding: A persistent queue imports GitHub issues or local PRDs, tracks states and priorities, enforces in-queue dependencies, skips duplicate IDs and processes items sequentially on the current branch.
-
Ralph for Claude Code docs/DOCKER_SANDBOX.md
Finding: Docker executes Claude inside a persistent container with a read-write project bind mount. Host orchestration remains local, CPU/memory/network are configurable, and credentials are scoped copies or API keys.
ralph-claude-code:TEAM-01: The product is a local CLI/terminal monitor without a shared multi-human prompting session.
ralph-claude-code:TEAM-01
The product is a local CLI/terminal monitor without a shared multi-human prompting session.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: visibility, discussion, comments, presence, ordering, conflicts, attribution, permissions, plan_restrictions, owner_online.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
ralph-claude-code:TEAM-02: Ralph has no product-owned human identity service, named team roles or enterprise SSO.
ralph-claude-code:TEAM-02
Ralph has no product-owned human identity service, named team roles or enterprise SSO.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: project_roles, task_roles, membership, provisioning, unauthorized_access, revocation, plan_scope.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
ralph-claude-code:TEAM-03: ALLOWED_TOOLS comes from repository-controlled configuration and is not a boundary against that repository. cross-checked
ralph-claude-code:TEAM-03
ALLOWED_TOOLS comes from repository-controlled configuration and is not a boundary against that repository.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code docs/CLI_OPTIONS.md
Finding: The CLI configures hourly calls, timeouts, tool allowlists, session resets and queue processing. Repository-controlled ALLOWED_TOOLS is explicitly not a security boundary.
ralph-claude-code:TEAM-04: Local queue state, logs and the last 50 session transitions persist on disk. cross-checked
ralph-claude-code:TEAM-04
Local queue state, logs and the last 50 session transitions persist on disk.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
-
Ralph for Claude Code docs/QUEUE_MANAGEMENT.md
Finding: A persistent queue imports GitHub issues or local PRDs, tracks states and priorities, enforces in-queue dependencies, skips duplicate IDs and processes items sequentially on the current branch.
ralph-claude-code:TEAM-05: Ralph monitor displays loop status and logs without proposed code changes or a code-diff review UI.
ralph-claude-code:TEAM-05
Ralph monitor displays loop status and logs without proposed code changes or a code-diff review UI.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: pending_reviews, evidence_navigation, comments, request_changes, shipping_guidance, effort, missed_evidence, keyboard, narrow_screen, context_switching, roles, retained_context.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
ralph-claude-code:SHIP-03: Queue processing commits one commit per issue on the current branch. With --github-issue and --create-pr, Ralph opens a pull request through the gh CLI on completion, optionally as a draft linked to the issue, and can comment on and close the issue. cross-checked
ralph-claude-code:SHIP-03
Queue processing commits one commit per issue on the current branch. With --github-issue and --create-pr, Ralph opens a pull request through the gh CLI on completion, optionally as a draft linked to the issue, and can comment on and close the issue.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code docs/QUEUE_MANAGEMENT.md
Finding: A persistent queue imports GitHub issues or local PRDs, tracks states and priorities, enforces in-queue dependencies, skips duplicate IDs and processes items sequentially on the current branch.
-
Ralph README: GitHub issue lifecycle
Finding: With --github-issue, Ralph uses the gh CLI to post progress comments, create a pull request on completion (optionally as a draft and linked with Closes #N), post a summary, close the issue and open follow-up issues.
ralph-claude-code:SHIP-05: --backup creates Git backup branches and --rollback restores a selected backup. cross-checked
ralph-claude-code:SHIP-05
--backup creates Git backup branches and --rollback restores a selected backup.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
ralph-claude-code:OWN-01: The distributed shell application source is MIT licensed. cross-checked
ralph-claude-code:OWN-01
The distributed shell application source is MIT licensed.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Finding: Ralph repository code is MIT licensed.
ralph-claude-code:OWN-02: The loop, queue and monitor run locally from the public source distribution. Claude authentication is required; optional E2B adds separate cloud credentials and compute costs. cross-checked
ralph-claude-code:OWN-02
The loop, queue and monitor run locally from the public source distribution. Claude authentication is required; optional E2B adds separate cloud credentials and compute costs.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
-
Ralph for Claude Code docs/E2B_SANDBOX.md
Finding: E2B runs Claude remotely while the host loop controls iterations and file sync. Sandbox expiry can recreate/reupload; cost limits use estimates rather than invoices.
ralph-claude-code:OWN-03: The inspected Ralph CLI implements no ADE analytics reporter or telemetry opt-out switch. Selected-agent, GitHub, sandbox and user-configured notification traffic remain separate.
ralph-claude-code:OWN-03
The inspected Ralph CLI implements no ADE analytics reporter or telemetry opt-out switch. Selected-agent, GitHub, sandbox and user-configured notification traffic remain separate.
- Documentation and pinned source inspection only; no hands-on scenario was run.
- Other atomic questions remain unestablished: code_data, prompts, credentials, logs, artifacts, retention, deletion, training, residency, subprocessors, telemetry_defaults, consent_prompt, provider_separation.
-
ralph-claude-code: ralph_loop.sh
Finding: The product installs Bash CLI commands and a display-only terminal/tmux monitor. Bounded inspection of the CLI, monitor and lib tree found no browser/phone application, multi-human prompting service, named workspace roles, enterprise SSO, code-diff review UI, agent-session delegation tools or schedule/event intake. Network integrations are the selected agent, GitHub operations, optional sandboxes and explicitly configured notifications. No ADE analytics reporter or telemetry opt-out switch is implemented; provider telemetry is separate.
ralph-claude-code:OWN-04: ralph-claude-code is MIT open source with no paid plan; it uses your own Claude credentials or Anthropic API key, and optional E2B sandboxes are billed by E2B (the tool only estimates and caps that cost). cross-checked
ralph-claude-code:OWN-04
ralph-claude-code is MIT open source with no paid plan; it uses your own Claude credentials or Anthropic API key, and optional E2B sandboxes are billed by E2B (the tool only estimates and caps that cost).
- First-party pricing pages, terms and licences read 2026-11-10; no purchase, billing or authenticated account test.
-
Finding: Says the tool is MIT licensed. Claude auth can use your own ANTHROPIC_API_KEY or Claude credentials. The optional E2B cloud sandbox has cost estimated from runtime times a user-set hourly rate, with budget caps, and E2B charges for it.
-
frankbria/ralph-claude-code LICENSE (MIT)
Finding: The MIT License lets anyone obtain, use, copy and modify the software free of charge.
ralph-claude-code:OWN-06: README reports active development at v0.11.5 with newer pinned-source changes described as unreleased. Public issues and contributor documentation are linked; no support response commitment is established. cross-checked
ralph-claude-code:OWN-06
README reports active development at v0.11.5 with newer pinned-source changes described as unreleased. Public issues and contributor documentation are linked; no support response commitment is established.
- Only the cited atomic findings are established. Remaining atomic requirements and the full rubric scenario remain unverified.
-
Ralph for Claude Code README.md
Finding: Ralph drives repeated Claude Code calls using local prompt/spec/task files. Product CLI commands, rate limits, circuit breakers, live output, session expiry, notifications and backup/rollback are documented.
ralph-claude-code:OWN-08: Ralph is a free MIT-licensed shell tool installed from the public repository; no paid plan is shown. Claude usage and optional E2B sandboxes are billed by their providers. cross-checked
ralph-claude-code:OWN-08
Ralph is a free MIT-licensed shell tool installed from the public repository; no paid plan is shown. Claude usage and optional E2B sandboxes are billed by their providers.
- Underlying Claude plan limits were not checked. No installed-product test was run.
-
Finding: Ralph is installed globally from a clone of the public repository with ./install.sh; the README names no price, account or paid plan.
-
Finding: The repository is under the MIT License, copyright Frank Bria.