agentleFS
CIrcui-tryCIrcui-try/Circuit/circuit_implementation_plan/AGENTS.md

Circuit / circuit_implementation_plan

AGENTS.md5 starsChanged 6 months agoMITView on GitHub

Circuit is not an IDE and not a code editor. Circuit is a visual editor for wiring local agent skills together. A skill is discovered only from: Assume the following are already implemented: Start with Phase 03 unless explicitly instructed otherwise. Do not mix these responsibilities: Frontend code must not directly depend on native host behavior. Use a bridge abstraction for: This bridge must be mockable in Playwright/UI tests. Native macOS folder pickers should not be directly automated in Playwright…

The file

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

10 rules. About 0 are common, 8 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    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
  1. Distinctive:

    Start with Phase 03 unless explicitly instructed otherwise.

    line 22
  2. Distinctive:

    Do not mix these responsibilities:

    line 28
  3. Distinctive:

    Do not add a code editor. Do not add global skill discovery.

    line 65
  4. Very specific:

    All Linear ticket lookups and creations must use the Circuit team space.

    line 70
  5. Distinctive:

    After completing a Phase, the coding agent must write a briefing in the format below.

    line 74
  6. Distinctive:

    Write the briefing in Korean (section headings may stay in English for template stability).

    line 76
  7. Very specific:

    Write the briefing as a file, not as a chat response.

    line 77
  8. Distinctive:

    Write the briefing only after all tests for that Phase pass.

    line 78
  9. Distinctive:

    After the briefing is written, all changes for that Phase must be recorded as a single commit.

    line 105
  10. Distinctive:

    Do not mix changes outside the Phase scope into the same commit.

    line 109

≈ “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

Coding Agent Guide

Project Intent

Circuit is not an IDE and not a code editor. Circuit is a visual editor for wiring local agent skills together.

A skill is discovered only from:

<repo>/.claude/skills/*/SKILL.md
<repo>/.codex/skills/*/SKILL.md

Current Project State

Assume the following are already implemented:

  • Phase 00: Tauri + React foundation
  • Phase 01: repository manager
  • Phase 02: skill discovery

Start with Phase 03 unless explicitly instructed otherwise.

Architecture Principles

1. Keep UI, Schema, and Runner Separate

Do not mix these responsibilities:

Visual Flow Editor
Workflow Schema
Manual Runner
Agent Adapter / Handoff

2. Use a Bridge for Host Capabilities

Frontend code must not directly depend on native host behavior. Use a bridge abstraction for:

  • repository selection
  • file system access
  • skill discovery
  • future command execution

This bridge must be mockable in Playwright/UI tests.

3. Do Not Automate Native File Dialogs in E2E

Native macOS folder pickers should not be directly automated in Playwright tests. Mock the bridge method instead.

4. Add Tests Phase by Phase

Starting from Phase 03, every phase must add or update at least one meaningful test.

Expected tooling:

Vitest for core logic
Playwright for UI/E2E

5. Preserve Product Scope

Do not add a code editor. Do not add global skill discovery. Do not add built-in default skills unless explicitly requested.

Linear

All Linear ticket lookups and creations must use the Circuit team space. Do not search or create tickets in any other team space.

Required End-of-Phase Briefing

After completing a Phase, the coding agent must write a briefing in the format below.

  • Write the briefing in Korean (section headings may stay in English for template stability).
  • Write the briefing as a file, not as a chat response. The path follows the pattern circuit_implementation_plan/phases/0N-{phase-slug}-briefing.md (e.g. 02-skill-discovery-briefing.md).
  • Write the briefing only after all tests for that Phase pass.
# Phase N Briefing

## Implemented
- Summarize the features that were implemented.

## Changed Files
- List the main files that changed and their roles.

## Verification
- Record the checklist you verified manually and how to run it.

## Tests
- List the tests you added or modified.
- Record the test commands and their results.

## Known Limitations
- List what is not yet implemented and what was intentionally left out.

## Next Recommendation
- Suggest what should be done in the next Phase.

Required Phase Commit

After the briefing is written, all changes for that Phase must be recorded as a single commit.

  • The commit subject must reference the Phase number (e.g. Phase 0: foundation, Phase 2: skill discovery).
  • The commit must only be made once the tests required by §"Add Tests Phase by Phase" are all green.
  • Do not mix changes outside the Phase scope into the same commit.

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.