loop-close
Closes a slice or a session - adversarial audit of the diff, ledger entry, selective commit, honest report, and the handoff for the next session. Use before…
Checks that the loop is actually installed and working in this project - files present, role profiles valid, connections alive, effort level real. Use after setup, after an upgrade, or when something behaves as if the method is not loaded.
$ npx -y skills add cbdreamer11/CB-loop-kit-claude-plugin --skill loop-doctor --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/loop-doctorContext preview
The summary Claude sees to decide when to auto-load this skill.
Checks that the loop is actually installed and working in this project - files present, role profiles valid, connections alive, effort level real. Use after setup, after an upgrade, or when something behaves as if the method is not loaded.
name: loop-doctor description: Checks that the loop is actually installed and working in this project - files present, role profiles valid, connections alive, effort level real. Use after setup, after an upgrade, or when something behaves as if the method is not loaded.
Verify the installation the same way the method verifies everything else: by observing, not by assuming it took.
`ls -la .loop/ .loop/roles/ .claude/agents/ 2>&1` and confirm: `STATE.md`, `VERIFY.md`, `DECISIONS.md`, `GOTCHAS.md`, the four role profiles, the `loop` wrapper (and that it is executable), and the role agents in the project's `.claude/agents/`.
Report anything missing, and whether `.loop/ACCESS.local.md` is gitignored — if it is tracked, stop and say so loudly: it is the file most likely to hold a credential.
`.loop/VERIFY.md` rots faster than anything else. **Run each declared command** and report which ones still work. A verification contract with a stale build command is worse than none, because it produces confident green nonsense.
Re-run the connection table from session 0: build, browser, database, payments test mode, git remote. PASS / FAIL / ABSENT. Every ABSENT should already be a declared GAP in `.loop/VERIFY.md`; if it is not, add it.
This session's effort level: **${CLAUDE_EFFORT}**
If that shows a level, that is the real one — including any silent downgrade. If it printed the placeholder literally, this version does not interpolate it; run `echo "$CLAUDE_EFFORT"` in a shell instead. There is deliberately no equivalent for the model: **no session can read which model it is running on**, so do not print a guess. If the role matters, relaunch with `./loop <role>` and trust the profile.
One table: what is healthy, what is broken with the exact command to fix it, and what is absent by design. If anything in section 1 or 2 is broken, say plainly that the method is **not** in force in this project yet.
A working method for building real software with coding agents, across many sessions.
Closes a slice or a session - adversarial audit of the diff, ledger entry, selective commit, honest report, and the handoff for the next session. Use before…
Sends an open design question to several independent voices in parallel and returns a decision with its evidence. Use before building anything that touches…
The planning session. Turns a goal into an ordered list of thin complete slices, decides which role and model runs each following session, and writes it all to…
Session 0. Sets up the loop in this project - detects the stack, checks which connections actually work (git, database, browser, payments sandbox), asks the…
Runs this project's verification contract and decides honestly whether something is verified, a gap, or a lie. Use before calling anything done, and whenever a…
The build loop - locate yourself, investigate the real code, build one slice completely, verify it for real, fix and re-verify, then close and hand off. Use…