Stats
Actions
Tags
From vcsdd
Reviews sprint contract in contracts/sprint-N.md using fresh vcsdd-adversary after Phase 2c, producing PASS/FAIL verdict and findings in reviews/contracts/sprint-N/output/.
How this command is triggered — by the user, by Claude, or both
Slash command
/vcsdd:vcsdd-contract-reviewThe summary Claude sees in its command listing — used to decide when to auto-load this command
## What Runs the sprint contract review gate. The adversary reviews the current sprint contract, the authoritative specs, and the current test/source artifact set to verify that the contract criteria are concrete, binary-evaluable, and aligned with what Phase 3 will judge. ## When Run after `/vcsdd-impl` completes Phase 2c and after the builder updates `contracts/sprint-N.md`. Requires the active feature to still be at phase `2c`. Required in `strict` mode before `/vcsdd-adversary`. Optional in `lean` mode when a sprint contract is still being used. ## How 1. **Resolve sprint number** fr...
Runs the sprint contract review gate. The adversary reviews the current sprint contract, the authoritative specs, and the current test/source artifact set to verify that the contract criteria are concrete, binary-evaluable, and aligned with what Phase 3 will judge.
Run after /vcsdd-impl completes Phase 2c and after the builder updates contracts/sprint-N.md. Requires the active feature to still be at phase 2c. Required in strict mode before /vcsdd-adversary. Optional in lean mode when a sprint contract is still being used.
state.json.sprintCountcontracts/sprint-N.md
vcsdd-contract.schema.jsondraft or under-review while negotiatingCRIT-XXXreviews/contracts/sprint-N/input/manifest.json:
contractDigest from the reviewed contract snapshot with line endings normalized and the status: frontmatter line ignored{
"reviewType": "contract",
"featureName": "...",
"sprintNumber": 1,
"contractPath": "contracts/sprint-1.md",
"contractDigest": "<sha256-of-reviewed-contract-with-status-normalized>",
"artifactsToReview": {
"spec": ["specs/behavioral-spec.md", "specs/verification-architecture.md"],
"tests": ["tests/..."],
"source": ["src/..."]
},
"reviewDimensions": ["spec_fidelity", "edge_case_coverage", "implementation_correctness", "structural_integrity", "verification_readiness"]
}
reviews/contracts/sprint-N/output/findings/reviews/contracts/sprint-N/output/verdict.jsonreviews/contracts/sprint-N/output/findings/FIND-NNN.jsonreviewContext.reviewType = "contract"reviewContext.contractPath = "contracts/sprint-N.md"reviewContext.contractDigest = manifest.contractDigestiteration = negotiationRound + 1PASS: the human may update contracts/sprint-N.md to status: approved
status field may change without rerunning reviewFAIL: revise the contract, increment negotiationRound, and rerunStrict-mode Phase 3 requires both:
contracts/sprint-N.md with status: approvedreviews/contracts/sprint-N/output/verdict.json with:
overallVerdict: "PASS"reviewContext.contractPathreviewContext.contractDigestiteration = negotiationRound + 1This prevents self-approved or post-review-edited sprint contracts from bypassing the adversarial gate.
/vcsdd-contract-review
/vcsdd-contract-review --sprint 2
npx claudepluginhub sc30gsw/vcsdd-claude-code --plugin vcsdd/70-phase-contractTransforms user task into verifiable contract.json: sets up contract directory via bash, launches contractor agent to scan codebase and assess risk, validates schema with retry.
/review-compactComprehensive review of Compact smart contract code covering 10 categories including privacy, security, tokens, concurrency, performance, and more, plus mechanical verification via /midnight-verify:verify. Supports parallel execution via agent teams (when enabled) or concurrent subagents.
/reviewAdversarially reviews a spec before build, citing evidence with file:line or source and ending with a GO/NO-GO gate.
/speckit.verifyValidates feature implementation by checking tasks, logic, tests, and code quality against specs. Produces verification-report.md.
/fire-4-verifyVerifies a completed phase by checking must-haves and running a 70-point quality validation checklist, then generates a detailed report and optionally routes to gap closure.
/evaluateRuns 3-stage verification (mechanical gates, semantic AC compliance, judgment quality) against seed spec. Blocks on stage 1 failures; run after /run before committing.