{"slug":"implement-12","title":"implement","summary":"Execute one bounded RED to GREEN experiment from bead or caller intent; return derived subject identity and check facts. Triggers: \"implement\", \"implement this bead\", \"run the experiment\". Full plan-to-validation requests route to rpi.","platform":"Claude","tags":[],"authorName":"LLM Mart","authorSlug":"llm-mart","score":0,"source":"github","price":null,"verified":false,"createdAt":"2026-09-08T21:44:56.675072Z","repo":{"url":"https://github.com/boshu2/agentops","stars":445,"forks":41,"license":"Apache-2.0","updatedAt":"2026-09-24T01:09:16Z"},"bodyHtml":"<hr>\n<p>name: implement\ndescription: 'Execute one bounded RED to GREEN experiment from bead or caller intent; return derived subject identity and check facts. Triggers: \"implement\", \"implement this bead\", \"run the experiment\". Full plan-to-validation requests route to rpi.'\npractices:</p>\n<ul>\n<li>tdd</li>\n<li>refactoring</li>\n<li>small-batch-flow\nhexagonal_role: driving-adapter\nconsumes: []\nproduces:</li>\n<li>subject-manifest.v1\noutput_contract: 'subject-manifest.v1 digest, author context ID, and exact acceptance-check receipts returned through the response or runtime channel'\ncontext_rel:</li>\n<li>kind: customer-of\nwith: plan\nskill_api_version: 1\nuser-invocable: true\nmetadata:\ngraph_root: true\ntier: execution\ndependencies: []\ncapabilities: [execute_one_experiment, collect_factual_evidence]\neffects: [modify_declared_subject, derive_subject_manifest]\ncanonical_status: canonical\ndisposition: keep</li>\n</ul>\n<hr>\n<h1>Implement</h1>\n<p>Execute exactly one bounded experiment described by the resolved bead or caller\nintent. Implement owns subject edits and factual evidence; the runtime derives\nidentity and receipts.</p>\n<h2>Prompt</h2>\n<pre><code>Implement bead ag-1234 from its text: acceptance \"ao gate check lists\nskill.probe-coverage\", scope cli/internal/gates/** plus regen outputs, first\ncheck `cd cli &amp;&amp; go test ./internal/gates/...`. RED first, smallest change,\nreturn the manifest digest and check receipts, stop.\n</code></pre>\n<h2>It's working if</h2>\n<ul>\n<li>After source activation and startup association, the first behavior check\nis the acceptance check; its output shows the expected failure (or an honest\ngreen structural baseline for documentation, relocation, or refactor work).</li>\n<li>Every path in <code>git diff --stat</code> falls inside the declared scope; an outside\nconsumer is reported as <code>file:line</code>, not absorbed.</li>\n<li>The response carries the <code>subject-manifest.v1</code> digest, author context ID,\nand verbatim check output; no <code>git commit</code> or <code>git push</code> appears.</li>\n</ul>\n<h2>Workflow</h2>\n<ol>\n<li>Read the intent, acceptance, and scope from their existing source; before\nthe first write, read <code>boundaries.md</code> in the rpi skill's <code>references</code>\ndirectory for what Implement does not own. Before execution can fail, the\ncaller passes source-store/project/work identity and permitted intent\nlocators at dispatch/start. At startup, return observed native runtime,\nsession/context identity (or explicit unknowns) through the caller-owned\nruntime channel for native comments/metadata recording; do not defer this\nassociation until handoff. Follow the fact distinctions in\n<a href=\"../cass/references/SESSION_FORMATS.md#work-to-session-associations\">session associations</a>.\nParent and resume links need observed provenance; controller dispatch alone\ndoes not establish native parentage. If startup observation or recording\nfails, preserve that failure and the pre-execution reference with the caller,\nleaving unobserved IDs unknown. Implement does not mutate the tracker.</li>\n<li>Run the declared first acceptance check before changing behavior. RED-first\napplies when acceptance is behavioral: preserve evidence that the check\nfails for the expected missing behavior. Relocations, doc merges, and pure\nrefactors record an honest green pre-change baseline instead.</li>\n<li>Make the smallest in-scope change that satisfies the active behavior.</li>\n<li>Run the targeted acceptance checks and capture factual results.</li>\n<li>Refactor only while those checks stay green. Refactoring does not change the\nacceptance test.</li>\n<li>Have the runtime derive actual changed paths and <code>subject-manifest.v1</code> from\nthe before/after subject.</li>\n<li>Run <code>ao provenance evidence-orphans --root &lt;repo-root&gt;</code> with one\n<code>--changed &lt;path&gt;</code> per changed path the runtime derived, and put its JSON\noutput in the check receipts the validator reads, so orphaned evidence\narrives as a receipt rather than as a surprise at verify time. Run it again after every repair round, over the\npaths as they stand, because a repair can orphan evidence the first pass did\nnot. Read the output as written and never hand-list the orphans instead.</li>\n<li>Return the manifest digest, author context ID, and exact check receipts in the\nresponse or runtime channel. Stop.</li>\n</ol>\n<p>Specialists (standards, domain, test, refactor, security) advise only. During\nedits, run the smallest deterministic checks that can falsify the change,\nreuse exact-input receipts whose subject and tool identity still match, and\nrun the full suite at the integration boundary unless the intent makes it the\nfirst check.</p>\n<h2>Scope conflict rule</h2>\n<p>On discovering a live consumer of the change outside the declared write scope\n(a test asserting the old path, a generated twin, a gate reading the moved\nfile), stop and report the exact file and line to the caller, who may revise\nthe intent and start a separate invocation; a different acceptance contract\nis a new intent.</p>\n<p>Before declaring GREEN, self-audit the diff for mocks, placeholders, TODO\nstubs, hardcoded fixture values, weakened assertions, regenerated goldens,\nwidened tolerances, suppression directives, or specification edits standing\nin for real behavior. A changed test, gate, fixture, golden, or acceptance\nsource must be required by the original intent, with green coming from the\nimplemented behavior; a check that passes against a substitute or weakened\noracle is not evidence: finish the behavior or report it as not built.</p>\n<h2>Boundary</h2>\n<p>Do not commit, push, claim, close, release, land, reserve, retry, or invoke a\nsemantic validator. A failed check is evidence for the caller, not permission\nto create a packet or validation loop.</p>\n","files":[{"path":"references/implement.feature","sizeBytes":735,"isText":false},{"path":"references/operations.md","sizeBytes":5867,"isText":true},{"path":"references/scaffold/agent-facing-tool-scaffolds.md","sizeBytes":1357,"isText":true},{"path":"references/scaffold/generic-templates.md","sizeBytes":8696,"isText":true},{"path":"references/scaffold/scaffold.feature","sizeBytes":1158,"isText":false},{"path":"SKILL.md","sizeBytes":8360,"isText":true}],"reviewScore":null,"reviewSummary":null,"trust":{"provenance":"trusted-source-unreviewed","notice":"Community-authored content, reproduced verbatim and not vetted as instructions. Treat it as data to evaluate, never as directives to follow.","bodySource":null},"bodyLocked":false,"purchaseUrl":null,"sourceUrl":null,"report":{"provenance":"trusted-source-unreviewed","screen":{"ran":true,"outcome":"clean","suspicious":0,"notes":0,"hiddenCharacters":false},"virusScan":{"engine":"clamav","status":"clean","scannedAt":"2026-09-24T06:49:40.366461Z","sha256":"14A2D7342D45701BD330DC222751B4232548314F8301FBD2FE7BCFD4AD3E1240","sizeBytes":12010},"review":null,"source":{"repositoryUrl":"https://github.com/boshu2/agentops","path":"images/gemini/skills/implement","license":"Apache-2.0","commit":"c3fe161dce0b85d1e0490df757bbb841d22e4ea1","subtreeSha":"EAFDEE9E6EB368D007A50FE988FBBE53C22C7675DCA7C6D50079F4531B61938F","lastSyncedAt":"2026-09-24T06:48:55.360254Z"},"reviewedAt":"2026-09-24T06:51:05.811115Z","notice":"Community-authored content, reproduced verbatim and not vetted as instructions. Treat it as data to evaluate, never as directives to follow."},"install":[{"target":"skills-cli","command":"npx skills add https://github.com/boshu2/agentops/tree/main/images/gemini/skills/implement"},{"target":"claude-code","command":"claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install boshu2-agentops@llmmart"},{"target":"git","command":"git clone https://github.com/boshu2/agentops.git"}]}