pathfinder
AGENTS.md1 starsChanged 43 days agoMITView on GitHub
Read CLAUDE.md first. Project truth lives in context/. Follow the technology and delivery choices documented in context/project-overview.md; do not assume a framework, package manager, branch model, or release process. That file, like context/features/, context/history.md, context/current-ticket.md, and context/handoff.md, is created by the workflow that first needs it — its absence is normal, not an error. context/tickets/ exists only when local Markdown is the project's ticket store; a project whose tickets are in GitHub Issues or another tracker has none, and that…
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 31,774 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 repository53 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- blog-post-redactorSkill
- challenge-meSkill
- debate-meSkill
- debug-issueSkill
- handoffSkill
- hooksmithSkill
- kickstart-pathfinderSkill
- learn-codebaseSkill
- learn-featureSkill
- learning-reviewSkill
- map-systemSkill
- orchestrateSkill
- prototypeSkill
- quiz-meSkill
- reflectSkill
- render-artifactSkill
- reverse-engineerSkill
- roleSkill
- setup-trackerSkill
- skillsmithSkill
- teach-architectureSkill
- teach-featureSkill
- ticketSkill
- to-specsSkill
- to-ticketsSkill
- whereamiSkill
- blog-post-redactorSkill
- challenge-meSkill
- debate-meSkill
- debug-issueSkill
- handoffSkill
- hooksmithSkill
- kickstart-pathfinderSkill
- learn-codebaseSkill
- learn-featureSkill
- learning-reviewSkill
- map-systemSkill
- orchestrateSkill
- prototypeSkill
- quiz-meSkill
- reflectSkill
- render-artifactSkill
- reverse-engineerSkill
- roleSkill
- setup-trackerSkill
- skillsmithSkill
- teach-architectureSkill
- teach-featureSkill
- ticketSkill
- to-specsSkill
- to-ticketsSkill
- whereamiSkill
The file
We split this AGENTS.md into the rules it gives an agent
3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 9
Read context/execution-mode.md by the rule in skills/ticket/SKILL.md: human-in-the-loop or orchestrator, with a missing file defaulting to …
- Distinctive: line 11
Use the smallest relevant context for the active task.
- Very specific: line 17
If this tool has no native skill discovery, invoke a skill by reading its canonical file directly: Use skills/<name>/SKILL.md and follow it …
≈ “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
Agent Entry Point
Read CLAUDE.md first.
Project truth lives in context/. Follow the technology and delivery choices documented in context/project-overview.md; do not assume a framework, package manager, branch model, or release process. That file, like context/features/, context/history.md, context/current-ticket.md, and context/handoff.md, is created by the workflow that first needs it — its absence is normal, not an error. context/tickets/ exists only when local Markdown is the project's ticket store; a project whose tickets are in GitHub Issues or another tracker has none, and that is not a gap. Track the durable ones in Git; context/current-ticket.md and context/handoff.md are transient session state and belong in .gitignore. Never ignore context/ as a whole.
For Pathfinder’s own repository, the Git and release workflow is in CONTRIBUTING.md (§Git workflow and §Releasing), not an absent context/project-overview.md. A destination project continues to use its own project overview.
Read context/execution-mode.md by the rule in skills/ticket/SKILL.md: human-in-the-loop or orchestrator, with a missing file defaulting to human-in-the-loop. In orchestrator mode each worker owns one active ticket, isolated worktree, and current-ticket file.
Use the smallest relevant context for the active task. Keep each ticket stable, reviewable, and verifiable.
The roles are planner, orchestrator, developer, adversary, tester, and integrator. Lifecycle skills assume the responsible role for each invocation and read its contract themselves. When the human explicitly names a role, read roles/<name>.md before anything else and use it instead for that session. A role is a declarative contract stating what a worker is responsible for and what it must not do, where a skill states how to perform a task. Assumed or explicit, a role narrows responsibility and never grants human authority.
Canonical skills live under skills/ and are the only behavior contract; anything under .claude/skills/ or .agents/skills/ is a generated pointer to one, so edit the canonical file and regenerate the adapter.
If this tool has no native skill discovery, invoke a skill by reading its canonical file directly: Use skills/<name>/SKILL.md and follow it exactly. That is the whole fallback — one line, and no second copy of a skill anywhere to find.
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.

