Conversation
…(issue #102 R11(B)) Lifts the seven universal sections from Sentinel's pre-DevTrail audit skill (audit/SKILL.md, contributed via issue #102) into a unified DevTrail v1 prompt template, parameterizing the project-specific hardcodes against the Charter doc / git range / originating AILOGs. This is the artifact PR — it ships the new template and schema update without yet touching the CLI resolver or removing the v0 templates. PR 4 will refactor the CLI to use this artifact and clean up legacy. NEW: dist/.devtrail/audit-prompts/audit-prompt.md (325 lines) Lifts integrally the seven universal sections that the issue #102 follow-up identified as portable: 1. ⛔ REGLA ABSOLUTA — SOLO LECTURA (read-only enforcement: "Si encuentras un bug, DOCUMENTALO. NO lo corrijas. Violación de esta regla invalida toda la auditoría.") 2. Tu rol — anti-cheerleader pattern. 3. Reglas de alcance — out-of-scope handled as separate notes, not as defects of the unit being audited. 4. Paso 2 verificación obligatoria — chain of tool-use evidence: locate → read FULL file → trace handler→service→repo→storage → read ≥2 test cases → cite path:line. Includes an explicit "Disciplina de evidencia" sub-block requiring tool-call-backed citations (R11(B) of issue #102 — paste-based audits without tool use produce structurally limited findings). 5. Paso 5 calibración severidad contra config REAL — anti-inflation and anti-deflation rules with the obligatory checklist (active driver, feature flags, build tags, DB role, deployment scope) before declaring Critical/High. Etapa 12 example (Pub/Sub stub vs gochannel active) preserved verbatim as a labeled real adopter case. 6. Lo que NO debes hacer — closing reinforcement (no modify, no inflate, no consult external sources). 7. Formato de salida — required frontmatter shape + body structure (Resumen, Verificación de compilación, Trazabilidad tarea por tarea, Hallazgos por severidad, Notas fuera de alcance, Evaluación de cierre, Conclusión). Sentinel hardcodes parameterized: - `specs/001-sentinel-mvp/tasks.md` → `{{charter_path}}` / `{{charter_content}}`. - `## Etapa N:` heading → `{{charter_id}}` / `{{charter_title}}`. - Project description "Sentinel es un monolito modular en Go..." → `{{project_context}}` (NEW placeholder, free-form, optional). - `internal/modules/identity/` and similar Go-specific paths → preserved as didactic illustrations (labeled as such), not parameterized — adopters in other stacks read them as patterns. - `go vet`, `go build`, `go test` commands → preserved in a block-quote labeled "Examples by stack" with parallels for Rust, TypeScript, and Python so non-Go adopters have a starting point. - `originating_ailogs[]` → `{{ailog_paths}}` + `{{ailog_contents}}` (DevTrail-specific, not in original Sentinel skill). Credit: explicit attribution at top (HTML comment header) and bottom (closing italics) to José Villaseñor Montfort (StrangeDaysTech) for the contributed source material via issue #102. UPDATE: dist/.devtrail/schemas/audit-output.schema.v0.json - audit_role enum extended to ["auditor", "auditor-primary", "auditor-secondary"]. v1 unified value is "auditor"; legacy v0 values still accepted during the v0→v1 transition. New audits should use "auditor". - NEW optional field `evidence_citations: integer (>=0)` — review-skill weighting hint that records how many path:line citations the auditor included. - calibratorOutput.auditors_reconciled.maxItems removed: v1 supports N≥2 auditors per cycle (previously fixed at exactly 2). - $comment updated with v1 evolution context. NEW: cli/tests/audit_template_test.rs (9 fixture tests) - unified_template_ships_at_canonical_path - unified_template_has_seven_universal_sections - unified_template_declares_expected_placeholders (incl. the new {{project_context}}) - unified_template_preserves_etapa_12_didactic_example - unified_template_credits_sentinel_contribution - unified_template_enforces_evidence_discipline - schema_accepts_v1_unified_audit_role (legacy + new coexist) - schema_declares_evidence_citations_optional - schema_calibrator_supports_n_auditors (no maxItems) NOT in this PR (deferred to PR 4): - DELETE auditor-primary.md, auditor-secondary.md, calibrator-reconciler.md (still present, still used by current CLI resolver — leaves PR 3 bisect-safe). - CLI resolver switch to read audit-prompt.md. - Canonical path migration (audit/charters/ → .devtrail/audits/). - Subcommand simplification (--prepare / --merge-reports). - Deprecation shims for --calibrate / --finalize. Test plan: - cargo test --test audit_template_test → 9/9 green - cargo test (full suite) → all passing, no regressions in the existing 14 charter_audit tests or the 8 audit_skill tests - No version bump (lands together with PRs 4-8 in the integrated v1 release per Propuesta/devtrail-audit-cli-flow.md v0.2 §5) Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This was referenced May 5, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Third of 8 PRs implementing the integrated v1 audit-skills iteration described in `Propuesta/devtrail-audit-cli-flow.md` v0.2. Closes the prompt content side of R11(B) of issue #102 — paste-based audits without tool use produce structurally limited findings; the new template enforces the discipline at prompt level.
Strategy: artifact PR, not refactor PR
This PR ships the new template + schema update without yet touching the CLI resolver or removing the v0 templates. PR 4 will do the refactor (switch resolver to read `audit-prompt.md`, delete legacy templates, migrate canonical paths to `.devtrail/audits/`, simplify subcommand). Splitting it this way keeps each PR bisect-safe and reviewable.
What ships
NEW `dist/.devtrail/audit-prompts/audit-prompt.md` (325 lines)
Lifts integrally the seven universal sections from Sentinel's pre-DevTrail `audit/SKILL.md` skill (contributed via issue #102):
Sentinel hardcodes parameterized
UPDATE `audit-output.schema.v0.json`
Credit
Explicit attribution to José Villaseñor Montfort (StrangeDaysTech) for the contributed source material at top (HTML comment header) and bottom (closing italics).
Tests
9 fixture tests in `cli/tests/audit_template_test.rs`:
NOT in this PR (deferred to PR 4)
Test plan
Phase v1 progress
🤖 Generated with Claude Code