メニュー
Freshly judge exact subject content against bead or caller acceptance, optionally persist verdict.v2 for a declared consumer, and stop. Triggers: "validate", "independently validate", "vibe".
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/boshu2/agentops --skill validateコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。
SKILL.md を表示中
Use Agent Mail as an optional messaging and file-reservation adapter for explicitly coordinated writers. Triggers: "coordinate writers", "reserve files".
Operate explicit orchestrator, implementer, validator, and scribe roles through a caller-selected agent runtime. Triggers: "agent-native factory", "role-shaped agent panes", "persistent workers".
Initialize minimal AgentOps documentation and verdict storage without taking over repository workflow. Triggers: "bootstrap AgentOps", "initialize AgentOps docs".
| name | validate |
| description | Freshly judge exact subject content against bead or caller acceptance, optionally persist verdict.v2 for a declared consumer, and stop. Triggers: "validate", "independently validate", "vibe". |
| practices | ["design-by-contract","llm-eval-harness","content-addressed-storage"] |
| hexagonal_role | driving-adapter |
| consumes | ["subject-manifest.v1"] |
| produces | ["subject-manifest.v1","validation-result","verdict.v2"] |
| context_rel | [{"kind":"customer-of","with":"plan"},{"kind":"customer-of","with":"implement"}] |
| skill_api_version | 1 |
| user-invocable | true |
| metadata | {"graph_root":true,"tier":"judgment","dependencies":[],"capabilities":["compute_subject_identity","judge_acceptance","return_validation_result","persist_verdict"],"effects":["write_verdict_artifact"],"canonical_status":"canonical","disposition":"keep"} |
| output_contract | PASS | FAIL | NOT_PROVEN with criteria, evidence, checked/not_checked, identity, and freshness; optional schemas/verdict.v2.schema.json persistence |
Independently judge one exact subject against the acceptance in its existing
bead or caller source, return one semantic result, and stop. Validate is the
sole verdict.v2 writer when persistence is requested. It never asks the model
to reconstruct Plan or Candidate packets.
store-verdict refuses an empty one. Plans, audits,
reviews, and other control artifacts are not completion subjects unless the
caller explicitly requested document review.source: runtime | caller and an
attester identity.Missing, colliding, or unattested identities produce NOT_PROVEN. This is a
declared trust fact, not cryptographic proof that contexts were isolated.
A caller may request that the fresh validator run on a different model than
the author. Dispatch via the controller-session recipe in
the agent-native model-dispatch recipe (codex-exec and/or ntm,
probed at runtime). Record author and validator model_identity in evidence
refs and freshness attestation notes — do not change verdict.v2 schema. If
the requested validator model has no live adapter, disclose the unsatisfied
diversity request and proceed same-model; never invoke claude -p /
claude --print. Single fresh validator remains the default shape.
Before running any acceptance-listed command, classify it as read-only or
subject-mutating. Regen scripts, sync scripts, formatters, and anything with
--force are subject-mutating until proven otherwise. Never run a
subject-mutating check against an uncommitted subject: on 2026-07-15,
scripts/test-ci-deterministic-gates.sh regenerated skills-codex/ from HEAD
mid-validation and destroyed the uncommitted subject, forcing NOT_PROVEN
(verdict b6e759dd...cb6a); only restoring the subject and revalidating in a
fresh context produced the PASS (e9b6cdb8...37b9). If a mutating check is
genuinely required by acceptance, run it against a disposable copy or a
committed subject, never the judged working tree.
not_checked has exactly one meaning: in-scope acceptance surface this
validation did not verify. PASS asserts that the whole declared acceptance
surface was verified, so a PASS carries no not_checked entries; the helper
refuses one and records a validate.integrity finding.
That rule never pays for deleting an honest caveat, because every kind of scope limit has a home that survives inside a PASS:
| Scope limit | Home | Example |
|---|---|---|
| A criterion proven by a bounded check | criteria[].reason on that criterion | "proven by the unit suite; the full integration matrix was not replayed" |
| A declared non-goal or out-of-scope area | the intent source's non-goals, optionally restated as an evidence-backed boundary criterion in criteria | "cli/** is a declared non-goal; the diff proves it untouched" |
| Residual risk or judgment caveat | the caller-facing report | "the migration path is untested against pre-3.0 stores" |
| Acceptance that genuinely went unverified | not_checked, and the result is NOT_PROVEN rather than PASS | "criterion 3 needs hardware this context cannot reach" |
Emptying not_checked to obtain PASS is a contract violation, not a
workaround. If acceptance really went unverified, the honest result is
NOT_PROVEN. If the entry was never acceptance in the first place, it belongs
in one of the other homes, where it stays visible in the stored artifact
instead of being deleted.
The helper ships beside this file. Invoke it through this skill's own
directory rather than a checkout-relative path: $SKILL_DIR is the directory
containing this SKILL.md — skills/validate/ in a repository checkout,
.agents/skills/validate/ in an installed runtime.
| Command | Required | Optional |
|---|---|---|
manifest | --root <dir>, --include <path> (repeatable, at least one) | --exclude <path-or-glob> (repeatable), --base-manifest <file>, --git-metadata-json <json>, --output <file> |
verify-manifest | --root <dir>, --manifest <file> | --base-manifest <file> |
snapshot-intent | --source <file> (- reads stdin) | --workspace <dir>, --intent-dir <dir> |
digest | <json-file> positional | none |
store-verdict | --draft, --intent-source, --subject-manifest, --author-context-id, --validator-context-id, --freshness-source <runtime|caller>, --freshness-attester-id, --scope-result <PASS|FAIL|NOT_PROVEN> | --workspace <dir>, --verdict-dir <dir> |
python3 "$SKILL_DIR/scripts/validate.py" manifest \
--root . --include skills/validate --exclude '**/*.log' --output manifest.json
subject-manifest.v1 with the manifest command
above (--root plus at least one --include). The helper uses only
filesystem content; Git commit/tree IDs are optional metadata. Derive the
manifest at the start of validation and re-derive it at the end; any
mismatch between the two is subject mutation and returns NOT_PROVEN.NOT_PROVEN.FAIL; incomplete scope evidence returns
NOT_PROVEN.FAIL, not
evidence of completion.
Judge every acceptance criterion and record criterion-level results,
findings, evidence references, checked, and any acceptance surface that
went unverified in not_checked (see Scope disclosure).PASS, FAIL, or NOT_PROVEN. Return
it with criterion results, findings, evidence references, checked,
not_checked, the acceptance and subject identities, distinct author and
validator context IDs, and the freshness attestation. PASS requires distinct
identities, explicit freshness, nonempty checked scope, top-level evidence,
evidence for every criterion, and an empty not_checked; route bounded
proofs, declared non-goals, and residual risk to the homes named in Scope
disclosure rather than deleting them or downgrading a proven result.verdict.v2 with the
helper's
store-verdict --draft <draft.json> --intent-source <resolved-intent> --subject-manifest <manifest.json> --author-context-id <id> --validator-context-id <id> --freshness-source <runtime|caller> --freshness-attester-id <id> --scope-result <PASS|FAIL|NOT_PROVEN>. The
helper snapshots the exact resolved intent under
<workspace>/.agents/ao/intents/sha256/<digest>.intent, then computes and
injects intent and subject digests plus author, validator, and freshness
facts. Identity and changed-path facts come from runtime-derived inputs and
receipts, not model transcription. Storage defaults to
<workspace>/.agents/ao/verdicts/sha256/<digest>.json; callers may provide
verdict_dir.The digest is SHA-256 over canonical JSON with artifact_digest omitted. Writes
use a same-directory temporary file, flush, fsync, and atomic rename. Identical
existing content is idempotent success; conflicting content is an integrity
failure represented by NOT_PROVEN.
Fresh validation means independent judgment over the exact subject. It does not require mechanically replaying every author command. Verify intent identity, scope, evidence digests, and every acceptance criterion; independently rerun the risk-critical, uncertain, or insufficiently evidenced checks. A digest-bound deterministic receipt may prove routine facts. Replay an expensive full suite only when acceptance requires that result or the supplied receipt cannot establish it.
Validate emits no WARN, confidence, disposition, briefing learning, owner, next action, repair, retry, replan, helper, escalation, tracker, Git, release, closure, or delivery state. Generic provenance may record a verdict later, but ledger availability cannot change its validity.