agentleFS
hranesshraness/atet/src/AGENTS.md

atet / src

AGENTS.md7 starsChanged 47 days agoMITView on GitHub

1 other project uses a version of this.

The file

We split this AGENTS.md into the rules it gives an agent

8 rules. About 2 are common, 3 are distinctive, and 3 are very specific to this project.

  • ≈2

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Keep canonical implementation in this source root and expose new behavior only through slopcamera commands, slopcamera.* operations, the …

    line 12
  2. Distinctive:

    Keep the declarative SDK and portable core canonical.

    line 14
  3. Distinctive:

    Treat explicitly imported workflow modules as trusted current-user Bun code.

    line 15
  4. Common:

    Keep portable source independent of repository-only packages and state, caller-selected executable code, arbitrary network URLs, and …

    line 16
  5. Very specific:

    Keep the ./workflow compatibility surface dependency-free.

    line 17
  6. Distinctive:

    Treat diagram and authored composition inputs as authoritative.

    line 19
  7. Very specific:

    Keep local vectorization authentication-free and network-silent.

    line 20
  8. Common:

    Keep this source root free of hosted-service concerns: no HTTP routes, product account, OAuth flow, credential persistence, or legacy …

    line 21

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

Contents

  • Canonical root modules – Slopcamera CLI, imperative workflow, diagram, canvas, artifact, direct Gateway generation, operation, MCP, and desktop-integration contracts.
  • icon.ts – the two-class product-identity pipeline: mark (bold compact favicon/app/header symbol) and illustration (bounded isometric marketing artwork) prompts, local ink extraction, VTracer trace, deterministic geometry gates, and purpose-specific vision-model critique rounds.
  • icon-scene.ts, soundtrack.ts, bounded-file.ts – the slopcamera.icon.* and slopcamera.soundtrack.* operations over the exact-version @hraness/iconplace and @hraness/soundfish libraries, with bounded reads and atomic replaceable writes.
  • code/ – canonical portable declarative graph authoring, closed public capability projection, compiler, plan, and execution contracts.
  • vectorize/ – bounded local raster decoding, VTracer supervision, SVG sanitization, measurement, provenance, and worker isolation.
  • *.test.ts and *.property.test.ts – deterministic examples, parser laws, and standalone consumer evidence.

Guidelines

  • Keep canonical implementation in this source root and expose new behavior only through slopcamera commands, slopcamera.* operations, the additive ./code graph SDK, and the compatible ./workflow SDK.
  • Preserve the root module and ./workflow v0.8 imperative APIs. Keep declarative authoring in ./code, lower-level compiler contracts in ./code/advanced, and testing or built-in workflow helpers outside the public export map.
  • Keep the declarative SDK and portable core canonical. The complete local host consumes that graph model and owns durable media execution. The Desktop shell supplies only native capture, permissions, and UI.
  • Treat explicitly imported workflow modules as trusted current-user Bun code. Compile each graph against one closed host projection and reject unsupported capabilities before executor or resource admission. Never add an open operation-registration hook.
  • Keep portable source independent of repository-only packages and state, caller-selected executable code, arbitrary network URLs, and ambient credentials.
  • Keep the ./workflow compatibility surface dependency-free. It may compose only the typed public operation registry, must parse foreign input, bound step dispatch, drain dispatched operations before returning, retain completed-step receipts on failure, and must not dynamically import authored code.
  • Parse every foreign value from unknown; bound source bytes, collections, dimensions, subprocess work, responses, and outputs before expensive or privileged work begins.
  • Treat diagram and authored composition inputs as authoritative; make rendered SVG, PNG, tldraw, and vector derivatives reproducible and replaceable.
  • Keep local vectorization authentication-free and network-silent. Send generation directly to the fixed Vercel AI Gateway origin using environment-only credentials, bounded responses, and maxRetries: 0.
  • Keep this source root free of hosted-service concerns: no HTTP routes, product account, OAuth flow, credential persistence, or legacy Graphics runtime surface. The hosted adapter under apps/api/ consumes these modules through the MCP runtime and owns all transport, billing, and storage state.
  • Pair parsing and compatibility changes with examples and property laws, then run the package check gate and clean standalone export.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.