- Home
- Skills
- Vladm3105
- Aidoc Flow Framework
- Doc Brd Audit
doc-brd-audit_skill
- Python
9
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
3 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill vladm3105/aidoc-flow-framework --skill doc-brd-audit- SKILL.md4.9 KB
Overview
This skill runs a single BRD audit workflow that executes a structural validator then a semantic reviewer and emits one combined audit report optimized for an automated fixer. It provides a clear pass/fail gate and a normalized fix queue so fix automation or manual triage can proceed reliably. Use it to simplify BRD quality checks in CI or pre-PR validation.
How this skill works
The skill runs doc-brd-validator first to check schema and template compliance, then doc-brd-reviewer to score content quality and alignment. Findings from both tools are normalized, merged, and formatted into a single BRD-NN.A_audit_report_vNNN.md containing summary, score calculation, validator and reviewer findings, diagram contract results, and a fix queue. If auto-fixable items exist, it can hand off the combined report to doc-brd-fixer.
When to use it
- You want a single user-facing command to run BRD quality gates
- You need a combined report as input to doc-brd-fixer
- Running CI or manual QA before PRD generation
- Validate both structure and content together before stakeholder review
- You need standardized fix metadata for automation
Best practices
- Run against the canonical BRD path (docs/01_BRD/BRD-NN_*/...) to ensure links and sections resolve
- Use the default reviewer threshold (90) or set a project-specific threshold explicitly
- Enable diagram contract checks when diagrams are required; use strict mode only when agreed across the team
- Ensure BRD files include recommended tags (e.g., @diagram: c4-l1, dfd-l0) and intent headers for better findings
- Review the Fix Queue output fields (source, code, severity, action_hint, confidence) before auto-fixing
Example use cases
- CI pipeline step that gates PRs containing BRD changes and blocks merges on FAIL
- Pre-release QA to produce a combined audit report for engineers and product owners
- Automated fix cycle: run audit, then hand report to doc-brd-fixer to apply auto_fixable fixes
- Manual triage: generate a single report for reviewers to assign follow-up work
- Project onboarding: validate a legacy BRD and generate a prioritized fix queue
FAQ
PASS requires validator PASS, reviewer score >= threshold (default 90), and no blocking issues. Any validator fail, reviewer score below threshold, or blocking/manual-required issue yields FAIL.
What format does doc-brd-fixer expect?
doc-brd-fixer accepts the combined BRD-NN.A_audit_report_vNNN.md (and legacy BRD-NN.R_review_report_vNNN.md). If both exist, the fixer should prefer the newest timestamp.