codomyrmex
AGENTS.md10 starsChanged 38 days agoMITView on GitHub
Status: Active | Last Updated: April 2026 This directory holds GitHub-specific configuration: CI/CD workflows, issue/PR templates, Dependabot config, CODEOWNERS, Gemini AI command configs, and multi-agent infrastructure. This document tells AI agents what they need to know to operate safely within this directory. Workflow YAML count: 37 files under .github/workflows/ (align with root README.md and docs/reference/inventory.md; refresh via uv run python scripts/doc_inventory.py).
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,700 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository47 other files it gives its agents.
- .agent/AGENTS.mdAGENTS.md
- .agents/AGENTS.mdAGENTS.md
- AGENTS.mdAGENTS.md
- .agents/skills/agent-interop/AGENTS.mdAGENTS.md
- .agents/skills/AGENTS.mdAGENTS.md
- .agents/skills/first-principles/AGENTS.mdAGENTS.md
- .agents/skills/mcp-tool-use/AGENTS.mdAGENTS.md
- .agents/skills/red-team/AGENTS.mdAGENTS.md
- .agents/skills/systems-thinking/AGENTS.mdAGENTS.md
- .codex/AGENTS.mdAGENTS.md
- config/AGENTS.mdAGENTS.md
- .cursor/skills/AGENTS.mdAGENTS.md
- .cursor/skills/codomyrmex/AGENTS.mdAGENTS.md
- .cursor/skills/desloppify/AGENTS.mdAGENTS.md
- .cursor/skills/fractals/AGENTS.mdAGENTS.md
- data/AGENTS.mdAGENTS.md
- .devcontainer/AGENTS.mdAGENTS.md
- docs/AGENTS.mdAGENTS.md
- evaluations/AGENTS.mdAGENTS.md
- examples/AGENTS.mdAGENTS.md
- .github/commands/AGENTS.mdAGENTS.md
- .github/ISSUE_TEMPLATE/AGENTS.mdAGENTS.md
- .github/scripts/AGENTS.mdAGENTS.md
- .github/workflows/AGENTS.mdAGENTS.md
- .jules/AGENTS.mdAGENTS.md
- .pipelines/AGENTS.mdAGENTS.md
- projects/AGENTS.mdAGENTS.md
- scripts/AGENTS.mdAGENTS.md
- src/AGENTS.mdAGENTS.md
- tests/AGENTS.mdAGENTS.md
- .workflows/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- agent-interopSkill
- first-principlesSkill
- mcp-tool-useSkill
- red-teamSkill
- systems-thinkingSkill
- agent-interopSkill
- first-principlesSkill
- mcp-tool-useSkill
- qmdSkill
- red-teamSkill
- systems-thinkingSkill
- CodomyrmexSkill
- desloppifySkill
- Fractals orchestratorSkill
- CodomyrmexSkill
The file
We split this AGENTS.md into the rules it gives an agent
9 rules. About 0 are common, 2 are distinctive, and 7 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
7
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 77
Read any file in .github/ to understand CI/CD structure and triggers
- Very specific: line 78
Run yamllint .github/workflows/*.yml to validate YAML syntax
- Very specific: line 79
Use gh run list --workflow=<name>.yml and gh run view to check status
- Very specific: line 80
Use gh pr list and gh issue list to review open items
- Very specific: line 81
Read CODEOWNERS to understand review assignment rules
- Distinctive: line 92
Read the target workflow file fully before editing
- Very specific: line 94
Keep concurrency: groups intact
- Very specific: line 96
Use uv (not pip/poetry) for Python dependency management
- Distinctive: line 97
Test workflow changes on a feature branch before merging to main
≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
.github -- AI Agent Guide
Status: Active | Last Updated: April 2026
Purpose
This directory holds GitHub-specific configuration: CI/CD workflows, issue/PR templates, Dependabot config, CODEOWNERS, Gemini AI command configs, and multi-agent infrastructure. This document tells AI agents what they need to know to operate safely within this directory.
Workflow YAML count: 37 files under .github/workflows/ (align with root README.md and docs/reference/inventory.md; refresh via uv run python scripts/doc_inventory.py).
Directory Contents
| Path | Purpose |
workflows/ci.yml | Continuous integration: lint, coverage-gate (full tests, 60% cov), Unix full-suite matrix, blocking Windows portability matrix, build |
workflows/pre-commit.yml | Pre-commit hook validation and commit message checks (soft-fail for agents) |
workflows/security.yml | Daily + push-triggered security scanning (6 scanners) |
workflows/release.yml | Tag-triggered release: quality gate (60% pytest cov), build, PyPI publish, stable-release documentation PR |
workflows/documentation.yml | Doc build and deploy on doc-file changes |
workflows/documentation-validation.yml | Link checking, structure validation (PR + weekly) |
workflows/benchmarks.yml | Performance regression tracking (push main + weekly) |
workflows/maintenance.yml | Stale issue/PR handling (90d for PRs, jules-exempt), cleanup, metrics |
workflows/workflow-coordinator.yml | Smart path-based selective workflow triggering |
workflows/workflow-status.yml | Aggregated status dashboard after workflow completions |
workflows/code-health.yml | Weekly code health dashboard with coverage trends |
workflows/auto-merge.yml | Squash-merge for PRs a reviewer labelled auto-merge (eligibility: scripts/maintenance/auto_merge_eligible.py) |
workflows/cleanup-branches.yml | Weekly cleanup of merged + stale branches |
workflows/pr-labeler.yml | Auto-labels PRs by paths, branch name, size, and module |
workflows/pr-conflict-check.yml | Scans open PRs for merge conflicts every 6 hours |
workflows/jules-dispatch.yml | Batch dispatch Jules tasks as GitHub issues |
workflows/agent-welcome.yml | Agent-specific PR welcome messages with checklists |
workflows/agent-metrics.yml | Weekly agent PR/issue/workflow health dashboard |
workflows/lock-threads.yml | Lock old closed issues (90d) and PRs (60d) |
workflows/gemini-*.yml | Gemini AI-assisted review, triage, dispatch (5 workflows) |
ISSUE_TEMPLATE/ | Structured issue forms: bug, feature, docs, Jules task |
commands/ | Gemini AI command .toml configs for PR review and triage |
CODEOWNERS | Code ownership mapping (@docxology) |
dependabot.yml | Automated dependency update configuration (3 ecosystems) |
PULL_REQUEST_TEMPLATE.md | Concise agent-friendly PR template (27 lines) |
Agent-Specific Infrastructure
Jules AI Agent
| Feature | Workflow | Details |
| Auto-labeling | pr-labeler.yml | Labels by branch name, paths, size, module |
| Auto-merge | auto-merge.yml | Squash-merge when checks pass + auto-merge label |
| Welcome message | agent-welcome.yml | Posts checklist comment on new Jules PRs |
| Batch dispatch | jules-dispatch.yml | Creates issues for each module from templates |
| Issue template | ISSUE_TEMPLATE/jules_task.yml | Pre-configured with all 28+ modules |
| Stale exemption | maintenance.yml | Jules PRs exempt from stale closure |
| Conflict detection | pr-conflict-check.yml | Auto-labels conflicting PRs every 6h |
| Metrics | agent-metrics.yml | Weekly dashboard of Jules PR/issue counts |
Gemini CLI Agent
| Feature | Workflow | Details |
| PR review | gemini-review.yml | Automated code review on PR open |
| Issue triage | gemini-triage.yml | Auto-triage on issue open/reopen |
| Dispatch | gemini-dispatch.yml | Routes @gemini-cli commands |
| Scheduled triage | gemini-scheduled-triage.yml | Periodic batch triage |
| Invocation | gemini-invoke.yml | Manual Gemini invocation |
Dependabot
| Feature | Details |
| Auto-labeling | pr-labeler.yml adds dependencies + automated (never auto-merge) |
| Auto-merge | auto-merge.yml merges when checks pass |
Safe Operations
- Read any file in
.github/to understand CI/CD structure and triggers - Run
yamllint .github/workflows/*.ymlto validate YAML syntax - Use
gh run list --workflow=<name>.ymlandgh run viewto check status - Use
gh pr listandgh issue listto review open items - Read
CODEOWNERSto understand review assignment rules
Caution Required
- Workflow edits take effect immediately on the next trigger — test on a branch
- CODEOWNERS changes affect review requests for every future PR
- Never remove
permissions: {}blocks — they enforce least-privilege tokens - Never remove
concurrency:blocks — they prevent duplicate workflow runs
Modifying Workflows — Checklist
- Read the target workflow file fully before editing
- Preserve all
permissions:blocks (deny-all default + job-level grants) - Keep
concurrency:groups intact - Pin action versions to a full commit SHA with a trailing comment naming the release, e.g.
uses: actions/checkout@<40-char sha> # actions/checkout@vX.Y.Z(CUR-4, CUR-5) - Use
uv(not pip/poetry) for Python dependency management - Test workflow changes on a feature branch before merging to main
- Validate YAML:
python3 -c "import yaml; yaml.safe_load(open('file.yml'))"
CI Quick Commands
# List recent CI runs
gh run list --workflow=ci.yml --limit=5
# Check auto-merge status
gh run list --workflow=auto-merge.yml --limit=3
# List agent PRs
gh pr list --label=jules --state=open
# List conflicting PRs
gh pr list --label=conflict --state=open
# View agent metrics (latest run)
gh run list --workflow=agent-metrics.yml --limit=1
# Trigger Jules batch dispatch
gh workflow run jules-dispatch.yml -f modules=all -f task_template=docstrings-mcp-testsOperating Contracts
| Contract | Requirement |
| Workflow top-level permissions | Must be permissions: {} (deny-all) |
| Job-level permissions | Grant only what the job needs |
| Action pinning | Full commit SHA with a trailing comment naming the release; never a tag, @main or @latest (CUR-4, CUR-5) |
| Concurrency | Every workflow must have a concurrency: block |
| Python tooling | Use uv for all dependency management |
| Test coverage | CI enforces 60% minimum (pyproject.toml) |
| Security scanning | 6 independent scanners must remain active |
| Agent PR handling | Auto-label → CI → auto-merge pipeline |
Currency and supply chain
Account rules CUR-1..CUR-16 are authoritative for this repository; where this file and those rules disagree, the rules win.
Navigation
- PAI.md | SPEC.md
- Root AGENTS | Root PAI
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

