# Chapter 8 — critic agent transcripts (captured)

Real output from the verification machine (same environment as the earlier transcripts). The Chapter 8 overlay (ch08-critic-agent) adds cognitionzoo/critic/critic.go (the agent), the workspace prompt file relurpify_cfg/prompts/agents/agent.critic.default.prompt, and cognitionzoo/critic/critic_agent_test.go.

# Package tests (ch02–ch08 overlays)

$ go test ./cognitionzoo/critic/... -count=1 -v        (32 tests)
--- PASS: TestCriticFailsClosedWithoutPermissionChecker (0.00s)
--- PASS: TestCritiqueWritesScratch (0.00s)
--- PASS: TestReviseFiresOnPredicateAndRecritiques (0.00s)
--- PASS: TestRevisionCapEndsLoopNormally (0.00s)
--- PASS: TestLibraryModeReportsWithoutRevision (0.00s)
--- PASS: TestCriticPromptFileLoadsAndValidates (0.00s)
--- PASS: (…the twenty-six Chapter 2–7 tests…)
PASS
ok  	codeburg.org/lexbit/relurpify/cognitionzoo/critic	0.061s

The six new tests, and what each pins:

• TestCriticFailsClosedWithoutPermissionChecker — deps without a permission checker: Execute returns critic.ErrNoPermissionChecker (asserted with errors.Is). The ReWOO bar, stated as an error rather than a silent downgrade.
• TestCritiqueWritesScratch — the draft phase runs through the delegate once, the critique phase issues one Chat with the authored criteria in the prompt (asserted via the fake model’s LastInput), the verdict lands in scratch.critic.verdict / scratch.critic.issues, and the result fields carry critique.
• TestReviseFiresOnPredicateAndRecritiques — the revise body runs exactly once when the verdict reads issues, then the loop re-critiques to a pass: two critique calls total.
• TestRevisionCapEndsLoopNormally — persistent issues run at most critic.MaxRevisionCycles (2) bodies, the telemetry sink records critic.revision_capped, and the run ends successfully with the last verdict standing — a cap is a bound, not a malfunction.
• TestLibraryModeReportsWithoutRevision — the zero value (no authored options) runs the library-mode graph: draft → critique → done, no revision.
• TestCriticPromptFileLoadsAndValidates — the workspace prompt file (shipped in the overlay at relurpify_cfg/prompts/agents/agent.critic.default.prompt, following the agent.<agent-name>.default convention) loads through the prompt registry from a fstest.MapFS and validates clean.

# Notes from the build

• The critic’s telemetry event strings follow the shipped <paradigm>.<event> vocabulary (critic.iteration, critic.reviewed, critic.revision_capped, critic.completed) as package-local constants; upstreaming named constants into the telemetry package’s vocabulary file is part of the Chapter 13 PR checklist.
• The prompt validator flags declared-but-unused variables (the shipped agent.react.default.prompt carries two as well); the critic’s file declares none — its criteria arrive through the agent’s options.
• The runner binary now links the full paradigm: contract, agent, prompt file, ensemble builder — and still runs its ensemble transcript unchanged.

# Gates

$ ./verify.sh gates ch08
== tier 2 (gates): ch08 ==
… (build, vet, tests, make lint-arch — all PASS)
[PASS] gates: ch08 holds the framework's law