letcairn.work
PassAudited by ClawScan on May 1, 2026.
Overview
The skill is a coherent markdown project-management workflow, but it asks users to install an external global CLI and creates persistent agent context files that should be reviewed before use.
This skill appears suitable for local markdown-based project management. Before installing, make sure you trust the cairn-work npm package, review the generated AGENTS.md and planning files, and use the execute autonomy level only when you are comfortable with the agent taking irreversible actions.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Installing the skill means trusting the external npm package that implements the CLI.
The skill relies on installing and running an external global npm CLI. That is expected for this CLI-based skill, but the install command is not version-pinned and the package implementation is not included in the provided artifacts.
npm install -g cairn-work cairn onboard
Install only from the expected npm package, consider pinning a known version, and review the package source or publisher before using it in sensitive projects.
Future agent sessions may reuse instructions or project context from these files.
The skill intentionally creates persistent agent-readable context files. This supports the project-management workflow, but those files can shape future agent behavior and may contain workspace-specific guidance.
`cairn onboard` creates `~/cairn/` with auto-generated context files (`AGENTS.md` and `.cairn/planning.md`) that agents read automatically.
Review AGENTS.md, .cairn/planning.md, and any workspace memory files periodically, especially after running update or onboarding commands.
If a task is marked execute, an agent may treat deployment, publishing, or sending as already approved.
The documented autonomy model can allow an agent to perform high-impact actions when a task is set to execute. The artifacts say the default is draft, so this is disclosed and user-controllable, but it deserves care.
| `execute` | Does everything, including deploy/publish/send | `completed` |
Keep tasks at propose or draft unless you intentionally want full autonomy, and reserve execute for tightly scoped, reversible work.