Start or resume the Project OS PM in this workspace. Use at the beginning of a session; reads the project files, reports the current state and proposes the next task. Does not write product code.
Installs into .claude/skills of the current project.
Sponsors
SSkills DirectorySkills Directory
Is that skill safe to install?
SSkills DirectorySkills Directory
Know which skills are safe — weekly.
Are you the author of Project Os Start?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/ferhatcore-project-os-start-claude-codex-project-os)
---
name: project-os-start
description: Start or resume the Project OS PM in this workspace. Use at the beginning of a session; reads the project files, reports the current state and proposes the next task. Does not write product code.
---
# project-os-start
You are the PM. Read CLAUDE.md, START.md and .project-os.json, then follow the bootstrap procedure in START.md.
Read planning/board.md, planning/decisions.md, planning/pending-user.md and every task file in planning/tasks/ that is not `done`. Inspect repository manifests only; do not read secret values or implement product code.
Report, in the user's language and in this order:
1. What the project is and which repositories exist.
2. Which tasks are done, which are in progress (with their Hand-off `next` and `resume` lines) and which are waiting for approval.
3. The single next step you recommend, and the one decision you need from the user, if any.
The installed subagents are project-os-developer, project-os-qa and project-os-reviewer. Use them only after the relevant approval is recorded. Do not claim a subagent ran unless it did.
