cfsa-antigravity 2.13.2 → 2.13.3

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "cfsa-antigravity",
-  "version": "2.13.2",
+  "version": "2.13.3",
   "description": "CFSA Pipeline — Constraint-First Specification Architecture for AI agents. Production-grade from line one.",
   "scripts": {
     "changeset": "changeset",

package/template/.agent/kit-sync.md

@@ -1,6 +1,6 @@
 # Kit Sync State
 
 upstream: https://github.com/RepairYourTech/cfsa-antigravity
-last_synced_commit: fa800ad9e35107aef8c6da0d2cc9161b379c71b2
-last_synced_at: 2026-03-20T01:43:15Z
-kit_version: 2.13.2
+last_synced_commit: 474ea4346ac09791e0f6bc7a5ed771de79ee93f2
+last_synced_at: 2026-03-20T14:04:21Z
+kit_version: 2.13.3

package/template/.agent/skills/pipeline-rubrics/references/scoring.md

@@ -19,9 +19,13 @@ Write all cross-layer findings to a `## Cross-Layer Consistency` section in the
 
 ## Document-to-Layer Mapping
 
+> **MANDATORY**: EVERY file matching these patterns MUST be individually audited through the full 3a→3b→3c cycle. Do NOT skip documents because rubric dimensions can be "mapped" to synthesized files. Do NOT read only "key documents" or "representative samples." The audit scope is EVERY file listed — no exceptions, no optimizations, no shortcuts.
+
+Use filesystem tools (`find_by_name`, `list_dir`) to discover ALL matching files. Write every discovered path to `audit-scope.md`.
+
 | Layer | Documents to load |
 |-------|-------------------|
-| Ideation | `docs/plans/ideation/ideation-index.md` + `ideation-cx.md` + all `*-index.md`, `*-cx.md`, and feature `.md` files recursively under `domains/` (and `surfaces/` for multi-product projects) |
+| Ideation | Use filesystem tools to recursively list ALL `.md` files under `docs/plans/ideation/`. Include every `*-index.md`, `*-cx.md`, feature `.md`, and `meta/*.md` file. For multi-surface projects, scan `surfaces/` subdirectories too. Every file found is audited — no selection, no sampling. |
 | Architecture | `docs/plans/*-architecture-design.md`, `docs/plans/ENGINEERING-STANDARDS.md` |
 | IA | `docs/plans/ia/index.md` + each shard listed + `docs/plans/ia/deep-dives/*.md` (list directory; include all files present) |
 | BE | `docs/plans/be/index.md` + each spec listed |

package/template/.agent/workflows/audit-ambiguity-execute.md

@@ -40,20 +40,24 @@ Read `docs/audits/audit-scope.md` and look for a `## Gaps Fixed` section.
 
 ---
 
+## 2.5. Coverage Counter
+
+Read `docs/audits/audit-scope.md`. Extract the `## Document Count` value and the full list from `## Documents to Audit`. State: **"This audit covers N documents. Every document will be processed through 3a→3b→3c. Documents processed: 0/N."**
+
+---
+
 ## 3a. Implementer Simulation
 
 Read .agent/skills/verification-before-completion/SKILL.md and follow its methodology.
 
-**CRITICAL ANTI-HALLUCINATION RULE**: You MUST NOT read all documents at once. You must process one document at a time through 3a → 3b → 3c before moving to the next document. Failure to follow this one-by-one sequence guarantees hallucinated citations and audit failure.
-
-*This is the most important step. It forces the agent to encounter the spec the way an implementer would, not the way a reviewer would.*
+**CRITICAL**: You MUST NOT read all documents at once. Process one document at a time through 3a→3b→3c. Do NOT skip documents because they seem redundant, because rubric dimensions "map" to other files, or because you can "derive" their content from synthesized documents. EVERY document in the scope list gets the full 3a→3b→3c treatment. After each document, state: **"Documents processed: X/N."**
 
 For each document:
 
-1. **Read** — Use a file reading tool to read the entire document end-to-end. Do not score yet.
-2. **Stub** — Attempt to write a stub implementation of each feature/endpoint/component using only what's in the spec — no external context or prior project knowledge.
-3. **Enumerate gaps** — List every decision you had to make that isn't explicitly specified: enum values, defaults, timeouts, error messages, retry counts, field types, HTTP status codes, validation rules, rate limit numbers, role permission lists, etc.
-4. **Write unconditionally** — Write each such decision immediately to `docs/audits/[layer]-ambiguity-report.md` as a punch list item with severity ❌.
+1. **Read** — Read the entire document end-to-end. Do not score yet.
+2. **Stub** — Write a stub implementation using only what's in the spec.
+3. **Enumerate gaps** — List every unspecified decision: enum values, defaults, timeouts, error messages, retry counts, field types, HTTP status codes, validation rules, rate limit numbers, role permission lists, etc.
+4. **Write unconditionally** — Write each gap immediately to `docs/audits/[layer]-ambiguity-report.md` as a punch list item with severity ❌.
 
 These gaps are **unconditional** — the rubric in 3b cannot override them.
 
@@ -83,6 +87,14 @@ After scoring all dimensions for a document:
 
 ---
 
+## 3d. Coverage Completeness Gate
+
+> **BLOCKING GATE — do not proceed until this passes.**
+
+Compare documents processed (from 3a counter) against documents listed in `audit-scope.md`. If `processed < listed`, list the skipped documents and STOP. Do NOT proceed to Step 3.5 or Step 4 with incomplete coverage.
+
+---
+
 ## 3.5. Cross-Layer Consistency Check
 
 This step runs after all per-document scoring is complete, whenever the audit scope includes BE, FE, or `all`.

package/template/.agent/workflows/audit-ambiguity-rubrics.md

@@ -31,13 +31,27 @@ Ask the user:
 - `fe` — Audit FE specs only
 - `all` — Full cascade (Vision → Architecture → IA → BE → FE)
 
-## 2. Load source documents
+## 2. Enumerate source documents
 
-Read `.agent/skills/pipeline-rubrics/references/scoring.md` for the document-to-layer mapping table. Load all documents for each layer being audited.
+Read `.agent/skills/pipeline-rubrics/references/scoring.md` for the document-to-layer mapping table.
 
-## 2.5. Persist audit scope
+**MANDATORY filesystem discovery**: Use file system tools (`find_by_name`, `list_dir`) to recursively discover ALL files matching the layer's patterns from `scoring.md`. Do NOT build the document list from memory or from reading an index file — the filesystem is the source of truth.
 
-Write `docs/audits/audit-scope.md` with the determined scope and document list:
+## 2a. Document Enumeration Gate
+
+> **BLOCKING GATE — do not proceed until this passes.**
+
+1. Count the total documents discovered by filesystem tools.
+2. State: **"Enumerated N documents for [layer] audit."**
+3. **Minimum counts** — if below these thresholds, re-scan:
+   - Ideation: count ALL `.md` files recursively under `docs/plans/ideation/` (expect 20+ for any real project)
+   - Architecture: ≥2 (`architecture-design.md` + `ENGINEERING-STANDARDS.md`)
+   - IA/BE/FE: ≥2 (index + at least one shard/spec)
+4. If re-scan still yields fewer than threshold, state why and proceed with the actual count.
+
+## 2b. Persist audit scope
+
+Write `docs/audits/audit-scope.md` with the determined scope and the **complete** document list from Step 2a:
 
 ```markdown
 # Audit Scope
@@ -48,7 +62,10 @@ Write `docs/audits/audit-scope.md` with the determined scope and document list:
 [list of selected layers]
 
 ## Documents to Audit
-[for each layer, list the exact file paths that will be audited]
+[for each layer, list EVERY file path discovered in Step 2a — no omissions]
+
+## Document Count
+[layer]: [N] documents
 
 ## Rubric Files
 [for each layer being audited, list the exact path of the rubric file]
@@ -62,7 +79,7 @@ Write `docs/audits/audit-scope.md` with the determined scope and document list:
 in-progress
 ```
 
-(Only include the rubric file entries for the layers actually being audited — not all five every time. This mirrors the conditional behaviour already applied to `## Documents to Audit`.)
+(Only include the rubric file entries for the layers actually being audited.)
 
 This file is read by `/audit-ambiguity-execute` as its prerequisite.