/cs-claude-rollout

Slash Command Source

Audit the Claude Code setup for $ARGUMENTS (default: the current repository) as a team policy.

Run

python ${CLAUDE_PLUGIN_ROOT}/skills/coding-agent-rollout/scripts/settings_linter.py --repo <repo> [--managed <managed-settings.json>]
python ${CLAUDE_PLUGIN_ROOT}/skills/coding-agent-rollout/scripts/mcp_plugin_inventory.py --repo <repo> [--allowlist <approved.json>]
python ${CLAUDE_PLUGIN_ROOT}/skills/coding-agent-rollout/scripts/claude_md_layering_audit.py --repo <repo>

If the user has adoption summaries, add:

python ${CLAUDE_PLUGIN_ROOT}/skills/coding-agent-rollout/scripts/adoption_tracker.py --merge <summaries...> --seats <N> [--weeks-since-launch <W>]

No managed settings yet? Start from skills/coding-agent-rollout/assets/managed-settings.example.json and lint it with --managed before proposing it.

Report

1. Blockers across all three scripts, each with its one-line fix.
2. Effective policy in plain words: what runs unprompted, what asks, what is denied, which MCP servers and plugins load.
3. Warnings grouped by script.
4. Adoption stage, if summaries were given.
5. Next phase from skills/coding-agent-rollout/assets/rollout_plan.md and its exit criterion.

Do not

• Do not edit settings files without being asked; propose the diff.
• Do not read or request session transcripts. Counts only.