本文へ移動
cccskills
無料GitHub で公開

articles-of-agreement-auditor

Audits an Articles of Agreement contract — the daemon-witnessed agreement every official Port Daddy agent signs covering registration, transcript reporting, tool-use gating, file claims, parley conduct, budget limits, and operator control — against the enforcement-beats-hope bar: a clause counts as safe only when it resolves to a concrete, daemon-observable mechanism with a defined denial shape, never a promise the agent might honor. Use when drafting a new agent's Articles, reviewing a compliance-level (C0-C6) claim, deciding whether a clause is genuinely enforced or merely documented, or auditing whether an agent's claimed identity can be trusted. NOT for securing the inbound event-to-spawn path (fleet-event-spawn-trust), designing the relay's transport PKI and signing keys (pd-relay-zero-trust), or tracking an agent's identity and reputation across its whole lifetime after it signs (agent-identity-continuity-reputation).

インストール方法を見る

含まれるファイル(11)

  • SKILL.md11.0 KB
  • agents/openai.yaml894 B
  • CHANGELOG.md218 B
  • examples/expected-output.md7.7 KB
  • examples/sample-input.json2.5 KB
  • README.md1.3 KB
  • references/compliance-levels-and-identity.md4.1 KB
  • references/enforcement-mechanism-taxonomy.md4.4 KB
  • schemas/articles.schema.json2.8 KB
  • scripts/articles_audit.mjs9.5 KB
  • templates/output-template.md2.0 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Articles of Agreement Auditor

Decide whether an Articles of Agreement contract actually binds an agent, or just documents good intentions.

Use This For

  • Drafting a new Agent Node's Articles before it registers: naming a concrete mechanism per obligation, not an aspiration.
  • Reviewing a claimed compliance level (C0 Registered through C6 Resumable) against the clauses that supposedly back it.
  • Deciding whether a specific clause is genuinely enforced (daemon can observe and, for gates, deny it) or merely written down.
  • Auditing whether a signing identity is trustworthy: daemon-issued at agent.register, and signed, not self-asserted by the body.
  • Catching a gate-style mechanism (pre-tool-gate, hook, capability-lease, mcp-gateway) with no defined denial shape before it ships.

Do Not Use This For

  • Securing the path from an inbound webhook/email/SMS/GitHub comment to an agent spawn — that is fleet-event-spawn-trust; this skill audits the signed contract an already-spawned agent operates under.
  • Designing the relay's transport PKI, per-publisher signing keys, or Merkle event chains — that is pd-relay-zero-trust; this skill is silent on wire-level trust between daemons.
  • Tracking a signed identity's behavior, outcomes, or reputation over its whole lifetime — that is agent-identity-continuity-reputation; this skill only checks the contract is sound and the identity was legitimately issued at signing time.

Enforcement Decision Model

flowchart TD
  A[Draft identity: daemonIssued + signed] --> B[Enumerate every obligation as a clause]
  B --> C[Assign an enforcementMechanism per clause]
  C --> D{Gate mechanism?}
  D -->|pre-tool-gate/hook/capability-lease/mcp-gateway| E[Define denialShape]
  D -->|probe/transcript-event| F[Leave denialShape unset]
  D -->|none| G[Honest but unenforced -- must fix before signing]
  E --> H[Run scripts/articles_audit.mjs]
  F --> H
  G --> H
  H --> I{pass:true and zero critical findings?}
  I -->|No| J[Fix the exact clause/identity field the finding names]
  J --> C
  I -->|Yes| K[Sign the Articles and register the Agent Node]
  1. Verify identity before evaluating anything else. Confirm identity.daemonIssued and identity.signed are both true. A body that can self-assert its own identity can also self-assert compliance — no clause matters until the signing identity itself is trustworthy.
  2. Enumerate every obligation as its own clause. Registration, transcript reporting, tool-use gating, file/symbol claims, parley conduct, budget/lease limits, and operator control each get a named clause — see templates/output-template.md.
  3. Assign a real enforcementMechanism per clause. Pick from pre-tool-gate, hook, capability-lease, mcp-gateway, probe, transcript-event, or the honest none — see references/enforcement-mechanism-taxonomy.md for which mechanism actually fits which obligation.
  4. Require daemonObservable: true for every clause that claims a mechanism. A named mechanism the daemon cannot independently confirm is functionally the same as none.
  5. Define denialShape for every gate-style clause. State the concrete artifact a violation produces (error code, refused call, revoked lease); never attach a denialShape to a passive probe/transcript-event clause, which has nothing to deny.
  6. Run scripts/articles_audit.mjs against the spec and fix every critical finding at the exact clause or identity field it names, not by weakening the check.
  7. Sign and register only on pass:true. A contract that fails closed is a blocker, the same as a red required CI check — not an FYI to route around.

Output Contract

The scorer reads a JSON object matching schemas/articles.schema.json:

  • identity: { daemonIssued: boolean, signed: boolean } — both must be true.
  • clauses[]: { name, obligation, enforcementMechanism, daemonObservable, denialShape? } — at least one clause; every clause's enforcementMechanism must not be "none" and daemonObservable must be true; every gate-style clause needs a non-empty denialShape.

Use scripts/articles_audit.mjs to audit an Articles spec JSON and return { pass, score, findings, recommendations }.

Anti-Patterns

The Hopeful Clause

Novice: Writes an obligation in plain English ("the agent will respect its budget") and calls it done, or picks a mechanism the daemon has no way to actually observe. Expert: Every clause resolves to a concrete, daemon-observable mechanism — or is honestly marked enforcementMechanism: "none" so the gap is visible, never silently assumed safe. Detection: articles_audit.mjs fires clause-not-enforced (critical) when a clause's enforcementMechanism is "none" or daemonObservable is false.

Self-Attested Identity

Novice: Lets the agent register with a self-chosen id, or treats an unsigned Articles document as close enough. Expert: The daemon mints the Agent Node id at agent.register against a nonce it issued first, and signs the Articles before any clause is trusted — a body can request capabilities, it cannot declare itself compliant. Detection: articles_audit.mjs fires identity-not-daemon-issued (critical) when identity.daemonIssued is not true, and identity-unsigned (critical) when identity.signed is not true.

The Gate With No Denial Receipt

Novice: Names a gate mechanism (pre-tool-gate, hook, capability-lease, mcp-gateway) and stops there, with nothing describing what actually happens on a violation. Expert: Every gate defines a denialShape — the literal artifact a violation produces (error code, refused call, revoked lease) — so the gate is provably built, not just named. Detection: articles_audit.mjs fires enforced-clause-no-denial-shape (critical) when a gate-style clause has no non-empty denialShape.

References

FileLoad When
references/enforcement-mechanism-taxonomy.mdNeed to pick a real enforcementMechanism for a clause, or decide whether it needs a denialShape.
references/compliance-levels-and-identity.mdNeed to ground a clause in the C0-C6 compliance ladder, or decide whether an identity claim can be trusted.
examples/expected-output.mdNeed to see a hopeful/self-attested contract audited, then the same contract fixed and passing.
examples/sample-input.jsonNeed a complete, passing Articles spec to copy and adapt.
templates/output-template.mdNeed a reusable Articles draft template (identity + clauses) to fill in before signing.
schemas/articles.schema.jsonNeed to validate an Articles spec JSON payload's structure before auditing it.
scripts/articles_audit.mjsNeed deterministic scoring of an Articles contract's enforceability.
agents/openai.yamlNeed a subagent descriptor for delegated Articles drafting/audit.
<!-- BEGIN BUNDLE INDEX (auto: index_references.py) -->

Skill Bundle Index

Every file in this skill, and when to open it. Auto-generated; run scripts/index_references.py --fix.

root

  • CHANGELOG.md — Articles of Agreement Auditor — Changelog — - Initial skill creation - Identity-then-clause enforcement decision model defined - Reference files and deterministic articles_audit script
  • README.md — Articles of Agreement Auditor — Audit an Articles of Agreement contract — the daemon-witnessed agreement every official Port Daddy agent signs — against the enforcement-bea

agents/

examples/

  • examples/expected-output.md — Example Output: Articles of Agreement Auditor — Scenario: a new Voyager is drafted with a self-issued identity, an unsigned Articles document, a "be nice" clause with no mechanism at all,
  • examples/sample-input.json — sample input (data/schema)

references/

schemas/

scripts/

templates/

<!-- END BUNDLE INDEX -->

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Expert in 2000s-era music visualization (Milkdrop, AVS, Geiss) and modern WebGL implementations. Specializes in Butterchurn integration, Web Audio API AnalyserNode FFT data, GLSL shaders for audio-reactive visuals, and psychedelic generative art. Activate on "Milkdrop", "music visualization", "WebGL visualizer", "Butterchurn", "audio reactive", "FFT visualization", "spectrum analyzer". NOT for simple bar charts/waveforms (use basic canvas), video editing, or non-audio visuals.

日本語の概要は準備中です。原文の説明を表示しています。

curiositech/port-daddy22026年10月12日 更新

Expert legal research agent for finding and scraping expungement data state by state. Knows authoritative sources, URL patterns, Firecrawl configuration, and 2026 legal landscape.

日本語の概要は準備中です。原文の説明を表示しています。

curiositech/port-daddy22026年10月12日 更新

Expert in 3D computer vision labeling tools, workflows, and AI-assisted annotation for LiDAR, point clouds, and sensor fusion. Covers SAM4D/Point-SAM, human-in-the-loop architectures, and vertical-specific training strategies. Activate on '3D labeling', 'point cloud annotation', 'LiDAR labeling', 'SAM 3D', 'SAM4D', 'sensor fusion annotation', '3D bounding box', 'semantic segmentation point cloud'. NOT for 2D image labeling (use clip-aware-embeddings), general ML training (use ml-engineer), video annotation without 3D (use computer-vision-pipeline), or VLM prompt engineering (use prompt-engineer).

日本語の概要は準備中です。原文の説明を表示しています。

curiositech/port-daddy22026年10月12日 更新

Apply crisis decision-making research to agent routing, uncertainty triage, and coordination failure analysis in time-pressured systems. Use when diagnosing handoff failures, analytical paralysis, or expert judgment under incomplete information. NOT for routine coding, simple CRUD design, or static single-agent tasks with complete information.

日本語の概要は準備中です。原文の説明を表示しています。

curiositech/port-daddy22026年10月12日 更新

Use for insight, reframing, contradiction, impasse, and anomaly-driven problem solving when execution effort no longer helps. NOT for routine optimization, error correction, or well-specified tasks with known solution paths.

日本語の概要は準備中です。原文の説明を表示しています。

curiositech/port-daddy22026年10月12日 更新

Apply cognitive task analysis to expert work that depends on perceptual cues, branching judgment, and recurring monitoring loops. Use when decomposing expert capability into agent structure, simulation design, or validation interviews. NOT for ordinary step-by-step SOP capture or simple pipelines with no tacit cue layer.

日本語の概要は準備中です。原文の説明を表示しています。

curiositech/port-daddy22026年10月12日 更新

curiositech のスキルをすべて見る

このスキルの問題を報告する