Pimp My IDE / Garage Dispatch
← Back to garage
September 18, 2026 · agent rules / executable proof / review systems

Gates beat good intentions.

Your AGENTS.md is a shop manual. Useful. Necessary. Also completely capable of being ignored by the same machine that just nodded at it. If a rule matters, give it teeth: a check that can turn the key off.

THE TAKE: Instructions shape behavior; gates constrain outcomes. The serious agent stacks are converging on the same move: map first, test claims, preserve rejected hypotheses, and make the final record machine-checkable.
Torque the release gate ↓

The instruction file is not the enforcement layer.

Gap-trap starts from an ugly, familiar failure mode: fast agent output makes humans stop reading every diff, while duplicate helpers, boundary violations, and tests-that-prove-nothing accumulate. Its answer is not a longer sermon. It generates repository-specific contracts, “proven red” checks that run new tests against old code, ratchets that allow known problems to fall but not rise, and playbooks for learned project facts.[1]

That split is the useful bit. The prose explains the rule. The gate catches the violation. A contract without a check is advice; a check without a readable contract is a land mine.

Words propose. Gates dispose.

Security work needs receipts for the misses, too.

Cloudflare’s security-audit skill describes a six-phase pipeline: reconnaissance, coverage-led hunting, candidate validation by a fresh verifier, structured findings, independent record verification, and reporting derived from those records. It keeps confirmed, needs_validation, and rejected verdicts distinct, and validates both the coverage ledger and findings records.[2]

That is better than a heroic agent dumping a confident report because it makes uncertainty inspectable. Rejected candidates are not wasted motion; they are proof that a path was tested and disproved. Coverage is not “the model seemed thorough.” It is an artifact another process can challenge.

The hybrid review stack is the grown-up answer.

Alibaba’s OpenCodeReview describes itself as a hybrid system: deterministic pipeline checks plus an LLM agent, with line-level findings and built-in rules for issues such as null-pointer errors, thread safety, XSS, and SQL injection. Its README also says it supports multiple model providers and several coding-agent surfaces.[3]

Do not read “hybrid” as boring compromise. It is division of labor. Deterministic machinery owns what can be stated as a repeatable invariant. Models own exploration, synthesis, and the weird semantic edges. Then an independent verification lane tries to break the model’s claim before the claim becomes a release decision.

Build the four-layer interlock.

Contract: name the invariant, scope, exceptions, and owner in human language. Executable gate: make the cheap, repeatable part fail locally or in CI. Adversarial verifier: ask a fresh process to disprove high-impact findings and record what it actually observed. Receipt: preserve inputs, revision, commands, results, unresolved facts, and the human decision.

The torque bench below replaces a single “AI review passed” badge with four native interlocks and a copyable release card. Closing every switch means the review package is structurally ready; it does not mean the code is correct, safe, or approved.

Interactive makeover / mechanical release interlock

Gate Stack Torque Bench.

Traditional purpose replaced: one green “agent reviewed” badge. Better version: four keyboard-native switches expose contract, executable proof, independent challenge, and release receipt as separate obligations.

Torque before merge

The gauge measures declared review structure, not correctness. Switches can close in any order because real retrofits rarely arrive neat.

Release interlocks

Release gate card

0/4 interlocks closed · RELEASE BLOCKED. Review structure is incomplete.

Open the three-source evidence drawer
[1] gap-trap README — contracts, proven-red tests, ratchets, playbooks, installation scope, and the project’s stated drift problem. [2] Cloudflare security-audit skill README — six-phase audit, coverage ledger, fresh candidate verification, structured verdicts, and independent final-record checks. [3] Alibaba OpenCodeReview README — project-authored description of hybrid deterministic/LLM review, line-level comments, built-in rule categories, providers, and supported surfaces.

Source boundary: all three sources are project-authored documentation. Feature and adoption claims are not independent benchmarks. The recommendation to combine readable contracts, executable gates, adversarial verification, and receipts is editorial synthesis.