panellisted

# /panel — methodology meta-review One invocation = one full meta-review of the agentic-ontology methodology. A single `methodology-reviewer` agent traces the whole current methodology end-to-end — `APPROACH.md`, the ADRs, the agent contracts, the skills, the playbooks, the case-law, and the live artefacts — diffs against the prior review, runs fresh blind spot-checks against the real target source, and emits real gaps + real improvement proposals. Per `APPROACH.md` §16. **History.** Through rev-8 this was the *Adversarial Review Panel* — six expert subagents + a chair, 7-13 agent invocations a run. Rev-9 (`retrospectives/LSN-024`) replaced the committee with one tracing review: the six were correlated Claude agents scoring conformance against a fixed target, with no memory; they re-listed stale findings every run and missed a whole methodology revision (rev-7's graph query layer). The `/panel` invocation is kept for continuity; it now runs one agent at roughly one-seventh the cost. Runs **periodically — per milestone, never per-commit** (cost discipline). Findings are candidates the maintainer triages; the review never auto-acts. ## Argument forms | Form | Behaviour | |---|---| | `/panel` | Run the methodology-reviewer — one pass. | | `/panel --repo <repo>` | Target a repo other than `odd-platform`. | | `/panel --show [<date>]` | Print the latest `review.md` (or the one for `<date>`). No agent spawned. | | `/panel validate` | Run the maintainer-authored acceptance-corpu