CRM
AGENTS.md989 starsChanged 6 days agoMITView on GitHub
Loaded by Codex every session. Other agents should follow the same files. Feature Proposals and RFC Reviews Discussions: docs/decisions/0003-rfc-community-prioritization.md. Read the AI operating model for task owners, handoffs and approval boundaries. Index: .agents/skills/churchcrm/SKILL.md. How to edit a skill: .agents/skills/churchcrm/skill-architecture.md. Load the card for the task. Do not load every skill. @.agents/skills/churchcrm/code-standards.md @.agents/skills/churchcrm/git-workflow.md Keep these heading names (this file links them): Commit freely, push only on approval, Mandatory Pre-Push Biome Check, Strict vs Loose Comparisons. Canonical setup and verification: DEVELOPING.md. Run…
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 31,439 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 repository10 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- churchcrmSkill
- source-command-church-reviewSkill
- source-command-pr-reviewSkill
- source-command-prune-branchesSkill
- source-command-push-prSkill
- source-command-repo-healthSkill
- source-command-security-advisorySkill
- source-command-security-report-triageSkill
The file
Show its rules and text
We split this AGENTS.md into the rules it gives an agent
13 rules. About 0 are common, 6 are distinctive, and 7 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
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
- could be automatic: a setting could enforce it
- Distinctive: line 11
Read the AI operating model for task owners, handoffs and approval boundaries.
- Distinctive: line 37
Keep these heading names (this file links them): Commit freely, push only on approval, Mandatory Pre-Push Biome Check, [Strict vs Loose
- Very specific: line 45
Run npm run env:check -- php before PHP work, or npm run env:check -- integration --services before integration tests.
- Very specific: line 49
Use the existing Compose stacks.
- Very specific: line 54
If a durable rule is missing or wrong, edit the existing card.
- Very specific: line 55
Do not append <!-- learned: --> essays.
- Distinctive: line 66
Do not land cross-cutting skill edits on an unrelated feature branch
- Very specific: line 67
After a push that fixes review comments, resolve every thread it fixed before you report done: check each against the pushed code, reply …
- Distinctive: line 68
Do not merge or close issues unless the maintainer answers yes to a direct question
- Very specific: line 71
Follow the human-first writing rules in docs/contributing/issue-readiness.md: BLUF first, concise goals and outcomes, details only when …
- Very specific: line 72
Write the current state: positive instructions, no history, no negating comments, migrate a legacy page when you touch it …
- Distinctive: line 73
Write little. Names and tests carry intent.
- Distinctive: line 74
If the PR author is the same GitHub user the agent is acting as, do not post a review comment.
≈ “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
ChurchCRM Development Guide
Loaded by Codex every session. Other agents should follow the same files.
Community feedback
Feature Proposals and RFC Reviews Discussions: docs/decisions/0003-rfc-community-prioritization.md.
Task ownership
Read the AI operating model for task owners, handoffs and approval boundaries.
Skills
Index: .agents/skills/churchcrm/SKILL.md.
How to edit a skill: .agents/skills/churchcrm/skill-architecture.md.
Load the card for the task. Do not load every skill.
| Task | Start |
| Review a PR | maintainer-review-gates.md → pr-review.md |
| Address review comments | .agents/skills/pr-review-fix.md → git-workflow.md |
| Commit / PR | git-workflow.md |
| UI | frontend-development.md |
| i18n | i18n-localization.md |
| Security | security-best-practices.md → authorization-security.md |
| Plugin | plugin-system.md → plugin-registry.md |
| Tests | ci-test-infrastructure.md → PHP tests or cypress-testing.md |
| API | api-development.md → service-layer.md |
Always-on:
@.agents/skills/churchcrm/code-standards.md
@.agents/skills/churchcrm/git-workflow.md
Keep these heading names (this file links them): Commit freely, push only on approval, Mandatory Pre-Push Biome Check, Strict vs Loose Comparisons.
Environment
Canonical setup and verification: DEVELOPING.md.
Run npm run env:check -- php before PHP work, or
npm run env:check -- integration --services before integration tests.
Prepare dependencies, browsers, and service images while network access is
available. The readiness check uses installed tools and never installs packages.
Use the existing Compose stacks; Playwright captures still require the separate
worktree/project described in marketing-visuals-pipeline.md.
Updating a skill
If a durable rule is missing or wrong, edit the existing card. Replace the sentence.
Do not append <!-- learned: --> essays. Do not paste catalogs from src/.
Plugin registry URL: CentralServices::PLUGIN_REGISTRY_URL (External branch today).
Issue and PR readiness
Before triaging or implementing an issue, read docs/contributing/issue-readiness.md. Check existing issues/PRs first, size work by review burden and risk, use epics for independently meaningful outcomes or cross-subsystem work, and treat good first issue as a maintainer certification. Keep issue intake friendly and PRs focused. Human maintainers remain accountable for decisions and approvals.
Session hygiene
- Confirm
git branch --show-currentbefore committing docs - Do not land cross-cutting skill edits on an unrelated feature branch
- After a push that fixes review comments, resolve every thread it fixed before you report done: check each against the pushed code, reply with the SHA when unclear, leave unfixed threads open (
git-workflow.md) - Do not merge or close issues unless the maintainer answers yes to a direct question
- User-visible behavior: open a docs tracking issue in ChurchCRM/docs.churchcrm.io (link it from the CRM issue when one exists). It does not block feature merge. Docs PRs wait for the release. See
docs/contributing/pr-review.md - UI that belongs on a marketing or docs page: one PR carries the code, the Cypress specs, and the Playwright capture spec (
playwright/workflows/;marketing-visuals-pipeline.md). Images, videos and the manifest are generated artifacts, never committed to this repo; visuals publishes them to cdn.churchcrm.io. Check captures in a git worktree on its own compose project, never on the dev checkout: the pipeline deletessrc/Include/Config.php. Open a separate Playwright issue only when the capture spec cannot be written yet - Follow the human-first writing rules in
docs/contributing/issue-readiness.md: BLUF first, concise goals and outcomes, details only when they help a decision. - Write the current state: positive instructions, no history, no negating comments, migrate a legacy page when you touch it (
skill-architecture.md) - Write little. Names and tests carry intent. Comment only to explain intent or a non-obvious constraint; do not narrate the code.
- If the PR author is the same GitHub user the agent is acting as, do not post a review comment. Fix the branch.
Tests
Run the failing spec in isolation before committing test fixes. Before a push, run the specs the change touches locally and check UI changes in a browser at desktop, tablet, and phone widths; UI PRs carry screenshots (git-workflow.md, cypress-testing.md). Do not assume Cypress / Yasumi fixture shapes — read the source. cy.visit() paths start with /.
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.

