Own whether a multi-document product-architecture binder is internally consistent, complete against its stated customer/contingency/architecture coverage, and honest about which older product ambitions it has absorbed, superseded, deferred, contradicted, orphaned, or rejected. Use when running a binder Architect of Record pass, before an implementation chain cites a binder chapter as ready, when a proof-gate owner is missing, or when reconciling the binder against the older ambition corpus (website, plans, examples, ADRs). NOT for auditing a single Claude Skill bundle's structure or frontmatter (use skill-hygiene), rendering a single-PM accept/reject verdict on one finished deliverable (use port-daddy-user-surrogate-pm-review), or sequencing/stewarding a roadmap's sidequests once a gap or operator decision has been surfaced (use legible-roadmap-with-sidequests).
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Une commande directe contourne le prompt de vérification. Examinez la source avant de l'exécuter.
Own whether a multi-document product-architecture binder is internally consistent, complete against its stated customer/contingency/architecture coverage, and honest about which older product ambitions it has absorbed, superseded, deferred, contradicted, orphaned, or rejected. Use when running a binder Architect of Record pass, before an implementation chain cites a binder chapter as ready, when a proof-gate owner is missing, or when reconciling the binder against the older ambition corpus (website, plans, examples, ADRs). NOT for auditing a single Claude Skill bundle's structure or frontmatter (use skill-hygiene), rendering a single-PM accept/reject verdict on one finished deliverable (use port-daddy-user-surrogate-pm-review), or sequencing/stewarding a roadmap's sidequests once a gap or operator decision has been surfaced (use legible-roadmap-with-sidequests).
license
Apache-2.0
allowed-tools
Read,Write,Edit,Bash,Grep,Glob
metadata
{"category":"Agent & Orchestration","tags":["binder-governance","architect-of-record","ambition-archaeology","raci-escalation","coverage-matrix"],"provenance":{"kind":"first-party","owners":["port-daddy"]},"pairs-with":[{"skill":"skill-architect","reason":"When a binder gap turns out to be a missing or malformed owning skill (no owner, no proof-gate mechanism), skill-architect designs or audits the skill that should close it."},{"skill":"product-reality-reviewer","reason":"Confirms whether a chapter's shipped-vs-target claim is actually true in running code; the Architect of Record cites that verdict as the evidenceLink behind a capability's proof."},{"skill":"port-daddy-user-surrogate-pm-review","reason":"Reviews a single finished deliverable's acceptance; the Architect of Record hands it a capability once owner+gate+evidence are all fixed, but does not itself render that accept/reject verdict."}],"io-contract":{"kind":"deliverable","consumes":["[Truncated]","[Truncated]"],"produces":["[Truncated]","[Truncated]"]}}
Architecture Binder Of Record
Own binder truth: is this multi-document product-architecture binder internally consistent, complete against its stated universe, and honest about shipped, partial, speculative, or blocked status — including the ambitions it inherited from before the binder existed.
Use This For
Deciding whether a chapter's claimed capability is real: owner assigned, acceptance gate named, evidence linked — not just confident prose.
Running a periodic ambition-archaeology sweep that classifies every old product promise (absorbed/superseded/deferred/contradicted/orphaned/rejected) instead of letting it rot unaddressed.
Tracking cross-chapter contradictions (term, authority, schema, shipped-vs-target) to resolution instead of letting two chapters quietly disagree.
Checking the three coverage axes (customer/deployment type, technical contingency, architecture consistency) for gaps before an implementation chain cites the binder as ready.
Writing the mandatory append-only binder-aor-log: ledger entry every run, including a run that finds nothing.
Do Not Use This For
Auditing a single Claude Skill bundle's structure, frontmatter, or file hygiene (skill-hygiene).
Rendering a single-PM accept/reject verdict on one finished deliverable (port-daddy-user-surrogate-pm-review).
Sequencing or stewarding a roadmap's sidequests once a gap or operator decision has been surfaced (legible-roadmap-with-sidequests).
Reconcile Loop
flowchart TD
A[Read last binder-aor-log entry] --> B[Scan changed chapters + source corpus since]
B --> C{Cross-chapter contradiction found?}
C -->|Yes, unresolved| D[Flag unresolved-contradiction, tier per severity]
C -->|No| E[Check each claimed capability]
D --> E
E --> F{owner + gate + evidence all present?}
F -->|No| G[Flag capability-without-owner-gate-evidence]
F -->|Yes| H[Run/continue ambition-archaeology sweep]
G --> H
H --> I{Every ambition classified?}
I -->|No| J[Flag ambition-unclassified]
I -->|Yes| K[Check the three coverage-matrix axes]
J --> K
K --> L{All 3 axes complete?}
L -->|No| M[Flag coverage-axis-incomplete]
L -->|Yes| N[Escalate per tier + write binder-aor-log entry]
M --> N
Read the previous binder-aor-log: note and reconcile from its window end, not a fixed time window — absence of a prior entry means start from the beginning.
Scan what changed since: binder chapters, open PRs touching runtime/binder surfaces, and recent notes mentioning the product's core nouns.
Classify every cross-chapter contradiction as , , , or , and mark it only once a source-linked fix has landed — never by assumption.
term
authority
schema
shipped-vs-target
resolved
Verify every claimed capability carries an accountable owner, a testable acceptance gate, and a link to real evidence (PR, transcript, screenshot, test run). Prose with no proof is not coverage.
Run or continue the ambition-archaeology sweep: classify every entry in the older ambition corpus (website, plans, examples, ADRs, idea troves) as absorbed, superseded, deferred, contradicted, orphaned, or rejected — see references/ambition-archaeology-classification.md.
Update the three coverage matrices (customer/deployment, technical contingency, architecture consistency) and escalate per tier — record, block a section, or request an operator decision. See references/raci-authority-and-escalation.md.
Write the mandatory binder-aor-log: ledger entry — confidence, findings, gates changed, handover — even when the run finds nothing. Absence of an entry is itself a finding for the next run.
Output Contract
A binder-coverage-spec that scripts/binder_coverage_audit.mjs reads carries:
documents[]: every binder document in scope, each with claimedCapabilities[] — { name, owner?, gate?, evidenceLink? }.
contradictions[]: every cross-chapter disagreement found — { kind: 'term'|'authority'|'schema'|'shipped-vs-target', resolved: boolean }.
ambitionCorpus[]: every older ambition family being reconciled — { name, classification: 'absorbed'|'superseded'|'deferred'|'contradicted'|'orphaned'|'rejected'|null }.
coverageMatrix: { customerAxisComplete, contingencyAxisComplete, architectureAxisComplete } — three booleans, each true only when every row on that axis has an owner, status, gate, failure mode, recovery path, and source.
Use scripts/binder_coverage_audit.mjs to audit a binder-coverage-spec JSON and return { pass, score, findings, recommendations }, including a binder-aor-log:-style summary line in recommendations on every run.
Anti-Patterns
Capability Claimed Without Proof
Novice: A chapter states "supports Work Intent as the sole launch primitive" with no owner, no acceptance gate, and no evidence link — the claim exists only as prose.
Expert: A claimed capability is not covered until it names an accountable owner, a testable acceptance gate, and a link to the evidence that proves it actually shipped and works.
Detection: binder_coverage_audit.mjs fires capability-without-owner-gate-evidence (critical) when any claimed capability is missing owner, gate, or evidenceLink.
Ambition Amnesia
Novice: An old public promise — a marketplace story, a pricing model, a named platform surface — quietly disappears from the corpus with no record of what happened to it.
Expert: Every entry in the ambition corpus gets one of six classifications (absorbed, superseded, deferred, contradicted, orphaned, rejected) with a rationale and a destination; "we forgot about it" is never a valid resting state.
Detection: binder_coverage_audit.mjs fires ambition-unclassified (critical) when any ambitionCorpus item has classification: null.
Coverage Theater
Novice: The binder is declared "complete" while two chapters still contradict each other, or a whole coverage axis (customer, contingency, or architecture) has never been checked.
Expert: Completeness is a positive, provable claim — zero unresolved contradictions AND all three coverage axes complete — never "nothing is currently flagged as broken."
Detection: binder_coverage_audit.mjs fires unresolved-contradiction (critical) for any contradiction with resolved:false, and coverage-axis-incomplete (critical) for any of customerAxisComplete / contingencyAxisComplete / architectureAxisComplete that is not true.
References
File
Load When
references/raci-authority-and-escalation.md
Need to know who is Accountable for a concern, what the Architect of Record may/may not do unilaterally, or which escalation tier a finding belongs to.
references/ambition-archaeology-classification.md
Need to classify an entry in the older ambition corpus, or need the required shape of a completed sweep.
examples/expected-output.md
Need to see an incomplete binder audited, then the same binder fixed and passing.
templates/output-template.md
Need a reusable run template: contradiction list, coverage matrix update, ambition table, and the mandatory ledger entry.
schemas/binder-coverage-spec.schema.json
Need to validate a binder-coverage-spec JSON payload's structure before auditing it.
scripts/binder_coverage_audit.mjs
Need deterministic scoring of a binder's completeness and consistency.
agents/openai.yaml
Need a subagent descriptor for a delegated Architect of Record run.
Skill Bundle Index
Every file in this skill, and when to open it. Auto-generated; run scripts/index_references.py --fix.
root
CHANGELOG.md — Architecture Binder Of Record — Changelog — - Initial skill creation - Reconcile-loop process, RACI/authority/escalation reference, and ambition-archaeology classification reference de
README.md — Architecture Binder Of Record — Own whether a multi-document product-architecture binder is internally consistent, complete against its stated coverage universe, and honest
examples/expected-output.md — Example Output: Architecture Binder Of Record — Scenario: a binder run claims a chapter's headline capability ("Work Intent is the sole launch primitive") with an owner but no acceptance g
references/ambition-archaeology-classification.md — Ambition Archaeology Classification — Use this when you need to classify an older public or internal product promise against the current binder, or when you're writing the ambiti
references/raci-authority-and-escalation.md — RACI, Authority, And Escalation — Use this when you need to know who is Accountable for a specific binder concern, what the Architect of Record may and may not do unilaterall
