aifolimizer
AGENTS.md2 starsChanged 5 months agoMITView on GitHub
Agent-optimized minimal context. Architecture, tech stack, MCP tool table, privacy rules, and code/workflow rules live in CLAUDE.md - read that first. This file lists only what's unique for an agent navigating the repo. MCP tool names use the mcp__aifolimizer__<tool> convention when invoked from Claude. Override with explicit path when triaging (specific log line, jsonl row).
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 31,542 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 repository29 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- adversarial-researchSkill
- auto-rebalanceSkill
- cash-deploymentSkill
- daily-briefingSkill
- dividend-strategySkill
- earnings-analyzerSkill
- earnings-postmortemSkill
- employer-stockSkill
- health-checkSkill
- macro-impactSkill
- momentum-scannerSkill
- optimize-allocationSkill
- pead-trackerSkill
- perf-optimizerSkill
- portfolio-healthSkill
- portfolio-reviewSkill
- position-reviewSkill
- pre-trade-checkSkill
- profile-setupSkill
- risk-assessmentSkill
- sector-rotationSkill
- stock-analysisSkill
- stock-compareSkill
- tax-loss-reviewSkill
- top-trades-todaySkill
- trade-journalSkill
- trading-deskSkill
- weekly-mirrorSkill
The file
We split this AGENTS.md into the rules it gives an agent
2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
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 61
Use ta lib for technicals - NOT pandas-ta (Python 3.14 incompatible.
- Very specific: line 63
Always call get_profile first in any analysis - never hardcode account types or capital.
≈ “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
aifolimizer - Agent Context
Agent-optimized minimal context. Architecture, tech stack, MCP tool table, privacy rules, and code/workflow rules live in CLAUDE.md - read that first. This file lists only what's unique for an agent navigating the repo.
Entry Points
backend/main.py- FastAPI appbackend/mcp_server.py- MCP server (107 tools)backend/mcp_login.py- interactive WS MFA login (run once)backend/run.py- uvicorn entry point.claude/skills/- 28 analysis skills
Important Folders
backend/app/services/- all data/compute logicbackend/app/api/ws.py- REST API routes.claude/context/- session state (changes.md, architecture.md, lessons.md, STATE.md)scripts/- automation (Telegram push, scheduled tasks, NSSM service)docs/- FAQ + sample skill outputs
Commands
# backend (Windows)
cd backend; .venv\Scripts\activate; uvicorn main:app --reload --port 8000
# register MCP server (one-time, absolute paths)
claude mcp add aifolimizer "<repo>/backend/.venv/Scripts/python.exe" "<repo>/backend/mcp_server.py"
# WS first-time login
cd backend; .venv\Scripts\python.exe mcp_login.py
# lint / compile / test
cd backend; ruff check .
cd backend; .venv\Scripts\python.exe -m py_compile mcp_server.py main.py
pytest # run from repo root (pyproject testpaths = backend/tests)
# infra (optional)
docker compose up -d# backend (macOS/Linux)
cd backend && source .venv/bin/activate && uvicorn main:app --reload --port 8000MCP tool names use the mcp__aifolimizer__<tool> convention when invoked from Claude.
Do Not Read
backend/.venv/backend/.pytest_tmp*/*.logbackend/data/*.jsonl(runtime data, not source).data/(Docker volumes - Postgres/Redis state)
Override with explicit path when triaging (specific log line, jsonl row).
Known Gotchas
- Use
talib for technicals - NOTpandas-ta(Python 3.14 incompatible; project pins 3.12). - MCP + FastAPI share L2 diskcache - cold MCP restart hits L2 if FastAPI warmed within TTL.
- Always call
get_profilefirst in any analysis - never hardcode account types or capital. - Single-letter
l(lowercase L) is ruff E741 - rename toloin candlestick code. .mcp.jsonat repo root is gitignored (contains absolute paths). Use.mcp.example.jsonas template..secrets/pg_password.txtmust exist beforedocker compose up- gitignored, create on first clone.
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.

