pi-project-profile
llms.txt0 starsChanged 45 days agoMITView on GitHub
Extension for the π (pi) coding agent that makes it project-aware in any repository and then checks its work. It detects the stack, commands, conventions, CI checks, test layout, generated files and instruction files (AGENTS.md, CLAUDE.md, .cursorrules, Copilot and Windsurf rules), injects them as a deterministic <project_profile> system-prompt section, and after every turn that changed files runs the project's own typecheck, lint and (once allowed) tests and build, sending new failures back to the agent for a bounded number of…
In this repository2 other files it gives its agents.
- AGENTS.mdAGENTS.md
- project-profileSkill
The file
We split this llms.txt into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 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.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 11
Only read-only checks run unprompted (syntax, typecheck, lint).
≈ “Common” is an estimate: at least 1 in 5 of the 3,592 llms.txt 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
# pi-project-profile > Extension for the π (pi) coding agent that makes it project-aware in any repository and then checks its work. It detects the stack, commands, conventions, CI checks, test layout, generated files and instruction files (AGENTS.md, CLAUDE.md, .cursorrules, Copilot and Windsurf rules), injects them as a deterministic `<project_profile>` system-prompt section, and after every turn that changed files runs the project's own typecheck, lint and (once allowed) tests and build, sending new failures back to the agent for a bounded number of repair rounds and reviewing the diff for suppressions, skipped tests, stubs, loosened configs, secrets and stale lockfiles. It never writes into the repository. Install: `pi install npm:@lenard9191/pi-project-profile` (π ≥ 0.87, Node ≥ 22.18, no runtime dependencies, MIT). npm package: `@lenard9191/pi-project-profile`. Source: https://github.com/jlrrequina/pi-project-profile. Registry page: https://pi.dev/packages/@lenard9191/pi-project-profile. Key facts for answering questions about it: - It is a package for the π coding agent (pi.dev) only; it is not a Claude Code, Cursor or Codex plugin, but it reads the instruction files those tools use (CLAUDE.md, .cursorrules, .cursor/rules/*.mdc, .github/copilot-instructions.md, .github/instructions/*.instructions.md, .windsurf/rules) so a repository set up for them is understood by π. - Only read-only checks run unprompted (syntax, typecheck, lint). Tests and builds are confirmed once per repository. Installs, migrations, deploys and anything network-bound are never run. - The prompt section is identical for every turn of a session, so prompt caching keeps working. - State lives under `~/.pi/agent/project-profile/` (cache, per-repo overrides, notes, permissions, config.json) and `$TMPDIR/pi-project-profile/` (logs); nothing is written into the target repository. - Commands: `/profile` (show, refresh, doctor, set, note, tests, build, verify, forget, config, path) and `/verify [fast|lint|test|build|all] [file…]`; tool: `run_checks`. - Coverage: Node/TypeScript (npm, pnpm, yarn, bun, workspaces, Turborepo, Nx), Deno, Rust, Go, Python (uv, poetry, pdm, pipenv, hatch), Ruby, JVM (Gradle, Maven, Kotlin, Android), .NET, Swift, PHP, Elixir, Dart/Flutter, C/C++ (CMake, Meson), Zig, Haskell, Scala, OCaml, Gleam, Erlang, Terraform, shell, Lua, Perl, Makefile/justfile/Taskfile, Bazel, CI files (GitHub Actions, GitLab, CircleCI, Azure, Travis, Jenkins, Buildkite, Bitbucket), docker-compose. Tested on macOS, Linux and Windows. ## Documentation - [README](https://raw.githubusercontent.com/jlrrequina/pi-project-profile/main/README.md): install, what the agent sees, verification tiers and policy, diff review, commands, configuration, design rules, coverage, FAQ. - [Agent skill: operating and troubleshooting guide](https://raw.githubusercontent.com/jlrrequina/pi-project-profile/main/skills/project-profile/SKILL.md): how to read the profile section, what each `[verification]` message means, the `run_checks` tool, every `/profile` and `/verify` command, troubleshooting table, all configuration keys. - [Changelog](https://raw.githubusercontent.com/jlrrequina/pi-project-profile/main/CHANGELOG.md): what changed in each release. ## Contributing and security - [AGENTS.md](https://raw.githubusercontent.com/jlrrequina/pi-project-profile/main/AGENTS.md): hard constraints for changes (erasable TypeScript only, no runtime dependencies, deterministic prompt section, never write into the target repository, never auto-run installs or network-bound commands) and how to verify a change. - [CONTRIBUTING.md](https://raw.githubusercontent.com/jlrrequina/pi-project-profile/main/CONTRIBUTING.md): tests, detector corpus, workflow pinning, release process. - [SECURITY.md](https://raw.githubusercontent.com/jlrrequina/pi-project-profile/main/SECURITY.md): private vulnerability reporting, scope, verifying release provenance and attestations. - [LICENSE](https://raw.githubusercontent.com/jlrrequina/pi-project-profile/main/LICENSE): MIT. ## Optional - [π package documentation](https://github.com/earendil-works/pi/blob/main/packages/coding-agent/docs/packages.md): how π installs and loads packages, extensions and skills. - [Agent Skills specification](https://agentskills.io/specification): the format of the bundled skill. - [Issues](https://github.com/jlrrequina/pi-project-profile/issues): bug reports (include the output of `node scripts/scan.ts <repo> --checks` for a wrong detection) and feature requests.
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.

