cococv
AGENTS.md181 starsChanged todayMITView on GitHub
- Installs packages
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 32,671 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 repository3 other files it gives its agents.
- web/AGENTS.mdAGENTS.md
- web/CLAUDE.mdCLAUDE.md
- resume-agentSkill
The file
We split this AGENTS.md into the rules it gives an agent
7 rules. About 0 are common, 4 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
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
- Very specific: line 19
No separate agent service on port 9100.
- Very specific: line 29
All modern client-side storage keys use cococv_* prefix (e.g.
- Very specific: line 56
Write added / fixed as short user-facing Chinese (no tech/ops internal details)
- Distinctive: line 58
Commit the changelog together with the release code, then push
- Distinctive: line 63
Do not modify unrelated files
- Distinctive: line 64
Read existing files and follow existing patterns before editing
- Distinctive: line 66
Do not move or rename files unless explicitly instructed
≈ “Common” is an estimate: at least 1 in 5 of the 41,581 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
AGENTS.md
Project Overview
- CocoCV (formerly Resume-Agent) monorepo with
frontend/andbackend/ - Backend is FastAPI served from
backend/main.py - Frontend is React 18 + Vite 5 + TypeScript in
frontend/ - Rebranded CLI supports dual binary entrypoints:
cococvandresume-agent
Documentation & Knowledge Architecture (Single Source of Truth)
- Public developer documentation lives in
docs/(tracked in Git; index atdocs/README.md):docs/frontend-architecture.md,docs/agent-streaming.md,docs/database-schema.md,docs/agent-glossary.md,docs/ai-ocr-and-pdf-cjk.md,docs/PI-CORE-MIGRATION.md, anddocs/adr/- Internal agent configs under
docs/agents/remain local-only (gitignored)
- Local
knowledge-base/is gitignored and holds only the lightweight archive index (knowledge-base/README.md)
Architecture Notes & Conventions (Milestone 2)
- Agent Runtime & Convergence:
- In-process Pi-Core single-coroutine linear event loop (
backend/agent/pi_core/) replacing legacy OpenManus inheritance chains - FastAPI backend architecture aligned with Strangler Fig evolution towards full-stack TypeScript
- No separate agent service on port 9100; agent runs in-process with main backend (
AGENT_BACKEND_BASE_URLdisabled by default)
- In-process Pi-Core single-coroutine linear event loop (
- LaTeX Compilation & Zero-Regression CJK /ToUnicode CMap:
- Both high-level
compile_latex(inbackend.latex_generatorandbackend.latex_compiler) and low-levelcompile_latex_raw(inbackend.latex_compiler) inject dual-layer CJK/ToUnicodeCMaps - Primary layer: fast-path via
pymupdf - Fallback layer: zero-dependency standard library incremental updater
_ensure_cjk_tounicode_cmaps_stdlib - Guarantees 100% extractable and searchable Chinese text in all generated XeLaTeX PDFs
- Both high-level
- Dual CLI Aliases:
pyproject.tomlregisters bothcococv = "backend.cli:main"andresume-agent = "backend.cli:main"- Both CLI entrypoints support identical subcommands:
init,schema,validate,render,sync, andskill
- Downward Storage Compatibility:
- All modern client-side storage keys use
cococv_*prefix (e.g.cococv_current,cococv_v2_data,cococv_resumes,cococv_pinned_ids) frontend/src/utils/storageCompat.tsprovidesgetCompatibleItem,setCompatibleItem, andremoveCompatibleItem- Reads check
cococv_*first; if absent, legacyresume_*/resume-agent-*keys are probed and automatically migrated forward tocococv_*
- All modern client-side storage keys use
Setup & Development
- Backend dependencies:
uv pip install -r requirements.txt(or symlinkedbackend/requirements.txt) - Frontend dependencies:
cd frontend && npm install
Run
- Backend:
python -m uvicorn backend.main:app --host 127.0.0.1 --port 9000 - Frontend:
cd frontend && npm run dev - CLI:
cococv --helporresume-agent --help
Testing & Quality Gates
- Frontend verification:
- Unit tests:
cd frontend && npm test -- --run - Production build:
cd frontend && npm run build
- Unit tests:
- Backend verification:
- Unit & regression tests:
pytest backend/tests/ - E2E tier tests:
pytest tests/e2e/tier1/
- Unit & regression tests:
- Rule: All testing must use deterministic native test runners (vitest, npm, pytest). Never invoke
claudeoropencodeCLI as test execution engines.
Engineering Governance & Release Process
- Deploy = manual deployment only (GitHub Actions set to
workflow_dispatch; serverauto-deploy.shcron paused). - Before pushing to main, update
CHANGELOGarray infrontend/src/data/changelog.ts:- Bump
version(patch +0.0.1, e.g.3.8.0→3.8.1; minor feature +0.1.0, e.g.3.8.0→3.9.0) - Date format must strictly be pure-date:
YYYY-MM-DDonly, NEVER include time (e.g.2026-11-15, forbidden:2026-11-15 10:00) - Write
added/fixedas short user-facing Chinese (no tech/ops internal details) - Multiple deploys on the same day merge into ONE entry (fold new items into today's entry and raise its version; do not append another entry)
- Commit the changelog together with the release code, then push
- Bump
- Users see a「有什么新变化」modal (
ChangelogModal) on first visit after each version bump.
Conventions
- Prefer minimal diffs
- Do not modify unrelated files
- Read existing files and follow existing patterns before editing
- Preserve user changes already present in the worktree
- Do not move or rename files unless explicitly instructed
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.

