Cross-repo conventions and audit system for AI-enhanced software projects.
24 standards defining convention rules for repos. 128 automated checks enforce them. Dashboard tracks compliance across all repos.
Built with AI assistance — see CREDITS.md.
standards/
├── *.md # Standard definitions (what + why)
├── scripts/
│ ├── audit.sh # CLI entry point — audit a single repo
│ ├── audit-lib.sh # Shared framework: _check, reporter, JSON output
│ ├── audit-all.sh # Batch audit across multiple repos
│ ├── agent-check.sh # Dispatches agent-eval requests for subjective checks
│ ├── checks/*.sh # One file per standard — plugins auto-discovered
│ └── ... # Utility scripts (badge generator, hooks, etc.)
└── .omo/dashboard/ # Generated audit results + HTML dashboard
- Standard documents (
*-standard.md) define each convention's rules, rationale, and success criteria. - Check scripts (
scripts/checks/*.sh) implement deterministic tests for each rule — file existence, grep patterns, format validation via shell. - Agent evals extend shell checks for subjective judgments (README quality, badge accessibility, SVG correctness). Checks write structured JSON prompts;
agent-check.shcollects them. - Audit runner (
audit.sh) discovers all check plugins, executes them against a repo, and produces terminal + JSON reports. - Dashboard (
audit-all.sh+dashboard.sh) runs batch audits across repos, aggregates into a compliance matrix, and generates a self-contained HTML dashboard.
# Audit the current directory
./scripts/audit.sh
# Audit + auto-fix (CREDITS.md, badges, README attribution)
./scripts/audit.sh --fix /path/to/repo
# Check a single standard, JSON output, exit 1 on failure
./scripts/audit.sh --standard ai-attribution --report json --exit-code /path/to/repo
# List all registered standards
./scripts/audit.sh --list-standards
# Batch audit all repos under a directory
./scripts/audit-all.sh -d /path/to/projects
# Generate dashboard from batch results
./scripts/dashboard.sh| Standard | File | Checks |
|---|---|---|
| ADR | adr-standard.md |
5 |
| AI Attribution | ai-attribution-standard.md |
6 |
| Auto-Commit GitOps | auto-commit-gitops-standard.md |
6 |
| Badge (shell) | badge-standard.md |
5 |
| Badge Quality (agent) | badge-standard.md |
3 |
| Changelog | changelog-standard.md |
6 |
| CI Pipeline | ci-pipeline-standard.md |
6 |
| Commit Conventions | commit-conventions-standard.md |
6 |
| CS Project Architecture | cs-project-architecture-standard.md |
4 |
| Dependency Management | dependency-management-standard.md |
8 |
| Git History Cleanup | git-history-cleanup-standard.md |
5 |
| GitHub Topics | github-topics-standard.md |
5 |
| .gitignore | gitignore-standard.md |
7 |
| Lefthook | auto-commit-gitops-standard.md |
6 |
| License | license-standard.md |
4 |
| Naming Conventions | naming-conventions-standard.md |
5 |
| Path Agnosticism | path-agnosticism-standard.md |
4 |
| README Quality | README-standard.md |
8 |
| Repo Structure | repo-structure-standard.md |
7 |
| Secrets Management (sops/age) | secrets-management-standard.md |
4 |
| SVG Screenshots | svg-screenshots-standard.md |
5 |
| Tool Versions (mise) | tool-versions-standard.md |
5 |
| Trivy Secrets | ci-pipeline-standard.md |
5 |
| Self-Consistency | self-consistency-standard.md |
1 |
Total: 24 standards, 128 checks (114 shell + 9 agent-pending + 5 tool-availability)
The audit runner returns exit code 1 if any check fails:
./scripts/audit.sh --exit-code /path/to/repo && echo "Compliant"Combine with --standard <name> for targeted CI gates.
Generate the compliance matrix across all repos:
./scripts/audit-all.sh -d /path/to/projects
./scripts/dashboard.sh
# Open .omo/dashboard/index.htmlThe dashboard shows per-repo, per-standard compliance with pass/fail/pending cells, a total score column, and a machine-readable compliance-matrix.json.
- Deterministic — same repo, same result. Shell checks before agent judgment.
- Plugin architecture — drop a
checks/<standard>.shfile, it auto-registers. - Agnostic — no hardcoded paths, no machine-specific config. Run from anywhere.
- Advisory by default — audit reports without modifying.
--fixfor remediation. - Exit-code gated —
--exit-codefor CI enforcement.
MIT — see LICENSE.