[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M759762F2G81N8PX96WPC18F":3,"skill-same:01M759762F2G81N8PX96WPC18F":50},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":26,"commitSha":27,"skillDirectory":28,"forksCount":29,"repositoryHtmlUrl":30,"repositoryHomepageUrl":31,"ownerHtmlUrl":32,"ownerBlogUrl":33,"ownerDescription":34,"updatedAt":35,"install":36,"resources":49},"01M759762F2G81N8PX96WPC18F","verification-contract",null,"Internal contract: the acceptance artifact per delivery unit — legacy (pre- feature 61) units may carry a frozen `ACCEPTANCE.md` (binding if present); post-feature 61 units use the unit doc's `## Acceptance criteria` section as the canonical acceptance source; the unit doc's evidence carrier — `## Evidence` table, or `## Log` `E` lines for a re-cut unit (#308) — serves as the execution receipt. Consumed by planners, execute-phase, and review-change.","gtrabanco\u002Fagentic-workflow","gtrabanco","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F1969593?v=4","01M7592KB3BVKCQFTXT94HVMT6","Gabriel Trabanco","01M7592KBTHE8HPQ8V87GJQYV8",23,"MIT","https:\u002F\u002Fgithub.com\u002Fgtrabanco\u002Fagentic-workflow\u002Fblob\u002Fmain\u002Fskills\u002Fverification-contract\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: verification-contract\nuser-invocable: false\nversion: 4.10.0\nauthor: \"Gabriel Trabanco \u003C1969593+gtrabanco@users.noreply.github.com>\"\nlicense: MIT\ndescription: >\n  Internal contract: the acceptance artifact per delivery unit — legacy (pre-\n  feature 61) units may carry a frozen `ACCEPTANCE.md` (binding if present);\n  post-feature 61 units use the unit doc's `## Acceptance criteria` section as\n  the canonical acceptance source; the unit doc's evidence carrier — `## Evidence`\n  table, or `## Log` `E` lines for a re-cut unit (#308) — serves as the execution\n  receipt. Consumed by planners, execute-phase, and review-change.\n---\n\n# Verification Contract (internal)\n\nSingle owner of the delivery finish line. Planning freezes it; executors may\nstrengthen coverage but cannot move it; review checks the same bytes.\n\n## Executable claims\n\nAn acceptance criterion or Evidence row that states a count, an exit code or a\nlength is a promise, not a description: `scripts\u002Fpromise-coherence.mjs` re-runs\nit. Write the outcome a command must produce (`` `cmd` matches \u003CN> ``,\n`` `cmd` exits \u003CN> ``) or, for a shape promise, name the file and the\n`## Section` the count is about (`` … a 3–5 line `## Inputs` in\n`skills\u002F\u003Cname>\u002FSKILL.md` ``) — a length with no named surface cannot be recounted\nand fails closed.\n\n## Artifact\n\nLegacy (pre-feature 61) units (ACCEPTANCE.md present — binding if present):\nthe frozen `ACCEPTANCE.md` beside the SPEC; required parts: `Status: frozen`;\none stable ID per SPEC criterion; `Required outcome`; named `Validator`; literal\nquality floor; project commands. Prefer commands, otherwise use\n`read-verified: \u003Cevidence>` or `manual: \u003Cexact observation>`. Unlabelled prose\nis invalid. A planned test may name its future project runner; it cannot\nsubstitute a narrower runner later.\n\nLane-era units (post-feature 61): the unit doc's `## Acceptance criteria` section is the\ncanonical acceptance source. Criteria are numbered, each induced from a concrete\nuser scenario, and command-verified where possible (label `read-verified` for\nprose-only checks). Unlabelled prose is invalid.\n\n**Every criterion names its demonstrating command (S5).** A criterion that a\ncommand can settle carries that command — ``\u003Ccmd> exits 0``, ``\u003Ccmd> matches\n\u003CN>`` — so a reviewer re-runs it instead of re-deriving the change, and `audit-pr`\ncan demand a demonstration for every AC. A criterion only a human can settle\nkeeps an explicit `manual:` label with the exact observation; a criterion with\nneither fails closed at review (the `no command, no claim` admission rule\napplied to the acceptance surface).\n\n**Validator stability.** A validator must never gate on a surface other workflow\nactors mutate — the branch diff as a whole, the session log, progress entries,\nreview ledgers, or forge state — because any out-of-unit commit (a session-log\nappend, another unit's fold) then re-fails a frozen criterion on a finished unit\nand re-opens its review loop. Grep the unit's own files and outputs; a\ndiff-based validator enumerates the unit's paths or excludes the\nworkflow-mutated surfaces explicitly (docs\u002FLOGS.md, the unit's own docs\ndirectory, harness\u002Ftoolstate).\n\n**Validator determinism.** A frozen validator must be deterministic — pin network,\nclock, randomness, and unordered-concurrency dependencies (fixed version,\nfrozen fixture, recorded input) or downgrade the row to `read-verified:`\u002F\n`manual:`.\n\n**Property \u002F fuzz validators.** A planner may emit a property or fuzz validator as\na bounded `commands` row — deterministic, bounded runtime within the project's\nper-command timeout budgets. No new validator kind in `VerificationPlan\nv1` is needed: property\u002Ffuzz rides the existing `commands` class.\n\n## Freeze and receipt\n\nThe unit doc's own evidence carrier is the execution receipt: the `## Evidence`\ntable for a legacy-shape unit, the `## Log` `E` lines for a re-cut one (#308). Each\nrow or line records a validator command, its exit code or blob digest, output\nexcerpt, and who verified it. A frozen criterion's receipt is the Evidence row for that\ncriterion — exact match continues; missing\u002Fmismatched evidence stops before\nedits:\n\n```text\nACCEPTANCE GATE — \u003Cunit> BLOCKED\nExpected Evidence row for AC\u003CN>: present with blob \u003Csha> · Actual: missing\u002Fchanged\nReason: the frozen finish line evidence is missing or changed.\n\n→ Next: restore the frozen manifest, or obtain explicit user approval for a\n  SPEC amendment and replacement manifest; then record a fresh Evidence row\n  · never edit tests, commands, or acceptance to make the current candidate pass\n```\n\nA legitimate change requires, in order: explicit user approval; dated SPEC\n`## Amendments` row; replacement manifest; committed fresh Evidence row. The\nexecutor never self-authorizes it.\n\nLegacy unit with no manifest mention: fingerprint committed `SPEC.md` and record\n`Manifest: legacy SPEC.md`. A new plan or any plan naming the manifest fails\nclosed when it is missing.\n\n## Which tests, and which layer\n\nThe acceptance artifact names the tests that judge a change; it never says \"the\nsuite\". A change derives its **test impact set** — the tests it can affect, from\nthe touched symbols, the paths its tests name, the suite beside each change and\nany LSP-resolved references (the block and its sources belong to\n`skills\u002Fexecute-phase\u002Freferences\u002FTEST_IMPACT.md`) — and that set plus the\nproject gate is what the receipt records.\n\nTwo verdicts, one receipt:\n\n- **targeted** — the set is non-empty and every criterion points at the tests in\n  it. A whole-gate run standing in for the set is a weaker receipt, not a\n  stronger one: it does not say what judged the change.\n- **empty** — only prose and assets changed, so no test can be affected. The\n  recorded empty verdict *is* the evidence for that phase. It is never a missing\n  row, and it never justifies an invented test.\n\nThe layer follows the change, not the runner: a plan\u002Fspec-only unit's evidence is\nits document checks; a domain or orchestration change's evidence is its derived\ntests. A phase that cannot name the tests it ran has not recorded a receipt, and\na phase that ran everything has not recorded which tests judge the change.\n\n## Validation ladder\n\nEvaluate every row plus the normal project gate:\n\n- `PASS`: commands green; read evidence present; manual checks named.\n- `FAIL`: validator disproves the candidate; include compact failure evidence.\n- `NEEDS-DECISION`: missing product\u002Farchitecture choice.\n- `BLOCKED`: command\u002Finput\u002Fenvironment unavailable; name it.\n\n## Anti-gaming rules\n\nForbidden: deleting, skipping, narrowing, or loosening a validator; suppression,\nstub, hard-coded answer, or no-op fix used to manufacture green. A command cannot\nprove an untested read\u002Fmanual row. Stronger regression tests are allowed. Repair\ntest setup only when assertions stay at least as strong and the reason is logged.\n\n## Test immutability\n\nA test, once written, is **immutable**. The executor fixes code until green,\nnever the test — editing an expectation to match behaviour is not a fix, it is\na cover-up. The sole legitimate amendment is a **proven mis-encoding of external semantics**: the test's expectation contradicts the actual\ndocumented semantics of the platform\u002Flibrary\u002Flanguage, cited from authoritative documentation — not a product decision change. Even that surfaces as a\n**finding plus a SPEC amendment**, never a silent edit to go green.\n\nThe deterministic backstop for this rule is the feature 60 path-protection\ncheckpoint (`path-protection@1` in the turn contract): the Tier 1 gate fails a\nprotected-path change with no recorded justification\u002Fapproval, and the pi guard\nblocks the write before it lands. This manifest is the finish line those\nrecorded escapes are checked against.\n\nPrevention rides the research gate (**research-before-encode**): platform\nsemantics are verified against authoritative documentation **before a test encodes them**, so **adding stronger tests stays allowed**; **editing expectations never** (except the proven-mis-encoding path above).\n\n## Done when\n\nExecution receipt (unit doc `## Log` `E` lines for a re-cut unit, `## Evidence` table for a post-feature 61 one; legacy `ACCEPTANCE.md` manifest binding-if-present) + named validators + literal quality floor; executor and reviewer evaluate identical evidence.\n","# Verification Contract (internal)\n\nSingle owner of the delivery finish line. Planning freezes it; executors may\nstrengthen coverage but cannot move it; review checks the same bytes.\n\n## Executable claims\n\nAn acceptance criterion or Evidence row that states a count, an exit code or a\nlength is a promise, not a description: `scripts\u002Fpromise-coherence.mjs` re-runs\nit. Write the outcome a command must produce (`` `cmd` matches \u003CN> ``,\n`` `cmd` exits \u003CN> ``) or, for a shape promise, name the file and the\n`## Section` the count is about (`` … a 3–5 line `## Inputs` in\n`skills\u002F\u003Cname>\u002FSKILL.md` ``) — a length with no named surface cannot be recounted\nand fails closed.\n\n## Artifact\n\nLegacy (pre-feature 61) units (ACCEPTANCE.md present — binding if present):\nthe frozen `ACCEPTANCE.md` beside the SPEC; required parts: `Status: frozen`;\none stable ID per SPEC criterion; `Required outcome`; named `Validator`; literal\nquality floor; project commands. Prefer commands, otherwise use\n`read-verified: \u003Cevidence>` or `manual: \u003Cexact observation>`. Unlabelled prose\nis invalid. A planned test may name its future project runner; it cannot\nsubstitute a narrower runner later.\n\nLane-era units (post-feature 61): the unit doc's `## Acceptance criteria` section is the\ncanonical acceptance source. Criteria are numbered, each induced from a concrete\nuser scenario, and command-verified where possible (label `read-verified` for\nprose-only checks). Unlabelled prose is invalid.\n\n**Every criterion names its demonstrating command (S5).** A criterion that a\ncommand can settle carries that command — ``\u003Ccmd> exits 0``, ``\u003Ccmd> matches\n\u003CN>`` — so a reviewer re-runs it instead of re-deriving the change, and `audit-pr`\ncan demand a demonstration for every AC. A criterion only a human can settle\nkeeps an explicit `manual:` label with the exact observation; a criterion with\nneither fails closed at review (the `no command, no claim` admission rule\napplied to the acceptance surface).\n\n**Validator stability.** A validator must never gate on a surface other workflow\nactors mutate — the branch diff as a whole, the session log, progress entries,\nreview ledgers, or forge state — because any out-of-unit commit (a session-log\nappend, another unit's fold) then re-fails a frozen criterion on a finished unit\nand re-opens its review loop. Grep the unit's own files and outputs; a\ndiff-based validator enumerates the unit's paths or excludes the\nworkflow-mutated surfaces explicitly (docs\u002FLOGS.md, the unit's own docs\ndirectory, harness\u002Ftoolstate).\n\n**Validator determinism.** A frozen validator must be deterministic — pin network,\nclock, randomness, and unordered-concurrency dependencies (fixed version,\nfrozen fixture, recorded input) or downgrade the row to `read-verified:`\u002F\n`manual:`.\n\n**Property \u002F fuzz validators.** A planner may emit a property or fuzz validator as\na bounded `commands` row — deterministic, bounded runtime within the project's\nper-command timeout budgets. No new validator kind in `VerificationPlan\nv1` is needed: property\u002Ffuzz rides the existing `commands` class.\n\n## Freeze and receipt\n\nThe unit doc's own evidence carrier is the execution receipt: the `## Evidence`\ntable for a legacy-shape unit, the `## Log` `E` lines for a re-cut one (#308). Each\nrow or line records a validator command, its exit code or blob digest, output\nexcerpt, and who verified it. A frozen criterion's receipt is the Evidence row for that\ncriterion — exact match continues; missing\u002Fmismatched evidence stops before\nedits:\n\n```text\nACCEPTANCE GATE — \u003Cunit> BLOCKED\nExpected Evidence row for AC\u003CN>: present with blob \u003Csha> · Actual: missing\u002Fchanged\nReason: the frozen finish line evidence is missing or changed.\n\n→ Next: restore the frozen manifest, or obtain explicit user approval for a\n  SPEC amendment and replacement manifest; then record a fresh Evidence row\n  · never edit tests, commands, or acceptance to make the current candidate pass\n```\n\nA legitimate change requires, in order: explicit user approval; dated SPEC\n`## Amendments` row; replacement manifest; committed fresh Evidence row. The\nexecutor never self-authorizes it.\n\nLegacy unit with no manifest mention: fingerprint committed `SPEC.md` and record\n`Manifest: legacy SPEC.md`. A new plan or any plan naming the manifest fails\nclosed when it is missing.\n\n## Which tests, and which layer\n\nThe acceptance artifact names the tests that judge a change; it never says \"the\nsuite\". A change derives its **test impact set** — the tests it can affect, from\nthe touched symbols, the paths its tests name, the suite beside each change and\nany LSP-resolved references (the block and its sources belong to\n`skills\u002Fexecute-phase\u002Freferences\u002FTEST_IMPACT.md`) — and that set plus the\nproject gate is what the receipt records.\n\nTwo verdicts, one receipt:\n\n- **targeted** — the set is non-empty and every criterion points at the tests in\n  it. A whole-gate run standing in for the set is a weaker receipt, not a\n  stronger one: it does not say what judged the change.\n- **empty** — only prose and assets changed, so no test can be affected. The\n  recorded empty verdict *is* the evidence for that phase. It is never a missing\n  row, and it never justifies an invented test.\n\nThe layer follows the change, not the runner: a plan\u002Fspec-only unit's evidence is\nits document checks; a domain or orchestration change's evidence is its derived\ntests. A phase that cannot name the tests it ran has not recorded a receipt, and\na phase that ran everything has not recorded which tests judge the change.\n\n## Validation ladder\n\nEvaluate every row plus the normal project gate:\n\n- `PASS`: commands green; read evidence present; manual checks named.\n- `FAIL`: validator disproves the candidate; include compact failure evidence.\n- `NEEDS-DECISION`: missing product\u002Farchitecture choice.\n- `BLOCKED`: command\u002Finput\u002Fenvironment unavailable; name it.\n\n## Anti-gaming rules\n\nForbidden: deleting, skipping, narrowing, or loosening a validator; suppression,\nstub, hard-coded answer, or no-op fix used to manufacture green. A command cannot\nprove an untested read\u002Fmanual row. Stronger regression tests are allowed. Repair\ntest setup only when assertions stay at least as strong and the reason is logged.\n\n## Test immutability\n\nA test, once written, is **immutable**. The executor fixes code until green,\nnever the test — editing an expectation to match behaviour is not a fix, it is\na cover-up. The sole legitimate amendment is a **proven mis-encoding of external semantics**: the test's expectation contradicts the actual\ndocumented semantics of the platform\u002Flibrary\u002Flanguage, cited from authoritative documentation — not a product decision change. Even that surfaces as a\n**finding plus a SPEC amendment**, never a silent edit to go green.\n\nThe deterministic backstop for this rule is the feature 60 path-protection\ncheckpoint (`path-protection@1` in the turn contract): the Tier 1 gate fails a\nprotected-path change with no recorded justification\u002Fapproval, and the pi guard\nblocks the write before it lands. This manifest is the finish line those\nrecorded escapes are checked against.\n\nPrevention rides the research gate (**research-before-encode**): platform\nsemantics are verified against authoritative documentation **before a test encodes them**, so **adding stronger tests stays allowed**; **editing expectations never** (except the proven-mis-encoding path above).\n\n## Done when\n\nExecution receipt (unit doc `## Log` `E` lines for a re-cut unit, `## Evidence` table for a post-feature 61 one; legacy `ACCEPTANCE.md` manifest binding-if-present) + named validators + literal quality floor; executor and reviewer evaluate identical evidence.\n",{"name":5,"author":24,"license":15,"version":25,"description":7,"user-invocable":20},"Gabriel Trabanco \u003C1969593+gtrabanco@users.noreply.github.com>","4.10.0","main","505c2244d42f95cdaac6dffd3c1c8c853545f9fb","skills\u002Fverification-contract",4,"https:\u002F\u002Fgithub.com\u002Fgtrabanco\u002Fagentic-workflow","https:\u002F\u002Fagwo.party","https:\u002F\u002Fgithub.com\u002Fgtrabanco","https:\u002F\u002Fgtrabanco.com","Sysadmin","2026-11-15 04:21:50",{"source":8,"skillName":5,"commands":37,"prompt":47,"zipAvailable":48},[38,41,44],{"tool":39,"command":40},"npx","npx skills add gtrabanco\u002Fagentic-workflow --skill verification-contract",{"tool":42,"command":43},"bunx","bunx skills add gtrabanco\u002Fagentic-workflow --skill verification-contract",{"tool":45,"command":46},"pnpm","pnpm dlx skills add gtrabanco\u002Fagentic-workflow --skill verification-contract","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fgtrabanco\u002Fagentic-workflow\u002Fblob\u002Fmain\u002Fskills\u002Fverification-contract\u002FSKILL.md",true,[],[51,59,66,73,80,87],{"skillNo":52,"name":53,"nameZh":6,"description":54,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":55,"categories":56,"keywords":57,"publishedAt":58},"01M7DY1H34KQK6218MKVHW1H4Z","runner-contract","The distributable runner's invocation contract: one train-stamped pin for every runtime command the workflow's skills cite. Referenced by the skills that run sensors, guards and ceremony operations; never a menu entry.","https:\u002F\u002Fgithub.com\u002Fgtrabanco\u002Fagentic-workflow\u002Fblob\u002Fmain\u002Fskills\u002Frunner-contract\u002FSKILL.md",[],[],"2026-11-15 14:19:17",{"skillNo":60,"name":61,"nameZh":6,"description":62,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":63,"categories":64,"keywords":65,"publishedAt":19},"01M7596P2HAKP84ZSMJWCPB0S5","skill-lint","Internal agentic-workflow maintenance: after SKILL.md edits, lint the repo's authoring rules, the machine-surface registration rules and the skill-content smell\u002Fsecurity checks. Triggers: \"lint the skill\", \"skill authoring check\", \"check the skill rules\", \"scan skills for smells\".","https:\u002F\u002Fgithub.com\u002Fgtrabanco\u002Fagentic-workflow\u002Fblob\u002Fmain\u002Fskills\u002Fskill-lint\u002FSKILL.md",[],[],{"skillNo":67,"name":68,"nameZh":6,"description":69,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":70,"categories":71,"keywords":72,"publishedAt":19},"01M7596DTAQ7CTTE64Q9YEMSDE","review-seo","Internal SEO review pass of the agentic-workflow review pack — composed in-turn by review-change and product-audit; not a menu entry. Checks changed web pages\u002Froutes for indexability, metadata, and structured data — applies only to public web surfaces. Findings only; never edits code.","https:\u002F\u002Fgithub.com\u002Fgtrabanco\u002Fagentic-workflow\u002Fblob\u002Fmain\u002Fskills\u002Freview-seo\u002FSKILL.md",[],[],{"skillNo":74,"name":75,"nameZh":6,"description":76,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":77,"categories":78,"keywords":79,"publishedAt":19},"01M75968HW51F81B96J1BWSDKX","review-security","Internal security review pass of the agentic-workflow review pack — composed in-turn by review-change and product-audit; not a menu entry. Checks secrets, input validation, injection, authn\u002Fauthz, PII exposure, and dependency risk on the changed surface. Findings only; never edits code.","https:\u002F\u002Fgithub.com\u002Fgtrabanco\u002Fagentic-workflow\u002Fblob\u002Fmain\u002Fskills\u002Freview-security\u002FSKILL.md",[],[],{"skillNo":81,"name":82,"nameZh":6,"description":83,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":84,"categories":85,"keywords":86,"publishedAt":19},"01M75964GC5PXNKNXVM4RPY081","review-perf","Internal performance review pass of the agentic-workflow review pack — composed in-turn by review-change and product-audit; not a menu entry. Checks the changed paths for algorithmic and resource regressions: N+1s, hot-path allocations, asset weight, and leaks. Findings only; never edits code.","https:\u002F\u002Fgithub.com\u002Fgtrabanco\u002Fagentic-workflow\u002Fblob\u002Fmain\u002Fskills\u002Freview-perf\u002FSKILL.md",[],[],{"skillNo":88,"name":89,"nameZh":6,"description":90,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":91,"categories":92,"keywords":93,"publishedAt":19},"01M7595V45AD7MDTC5WJKHGMTK","review-design","Internal UI\u002FUX design review pass of the agentic-workflow review pack — composed in-turn by review-change and product-audit; not a menu entry. Checks the changed UI against the project's design doc: consistency, states, responsiveness, and reuse — applies only when the project has a UI and the change touches it. Findings only; never edits code.","https:\u002F\u002Fgithub.com\u002Fgtrabanco\u002Fagentic-workflow\u002Fblob\u002Fmain\u002Fskills\u002Freview-design\u002FSKILL.md",[],[]]