The-Ouroboros/CLAUDE.md

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Git Workflow Policy

CRITICAL: All code changes MUST follow this workflow. Direct pushes to main are ABSOLUTELY PROHIBITED.

1. Create Gitea Issue First — All features, bug fixes, and policy changes require a Gitea issue before any code is written
2. Create Feature Branch — Branch from main using format feature/issue-{N}-{short-description}
3. Implement Changes — Write code, tests, and documentation on the feature branch
4. Create Pull Request — Submit PR to main branch referencing the issue number
5. Review & Merge — After approval, merge via PR (squash or merge commit)

Never commit directly to main. This policy applies to all changes, no exceptions.

Agent Workflow

Modern AI development leverages specialized agents for concurrent, efficient task execution.

Parallel Execution Strategy

Use git worktree or subagents (via the Task tool) to handle multiple requirements simultaneously:

• Each task runs in independent context
• Parallel branches for concurrent features
• Isolated test environments prevent interference
• Faster iteration with distributed workload

Specialized Agent Roles

Deploy task-specific agents as needed instead of handling everything in the main conversation:

• Conversational Agent (main) — Interface with user, coordinate other agents
• Ticket Management Agent — Create/update Gitea issues, track task status
• Design Agent — Architectural planning, RFC documents, API design
• Code Writing Agent — Implementation following specs
• Testing Agent — Write tests, verify coverage, run test suites
• Documentation Agent — Update docs, docstrings, CLAUDE.md, README
• Review Agent — Code review, lint checks, security audits
• Custom Agents — Created dynamically for specialized tasks (performance analysis, migration scripts, etc.)

When to Use Agents

Prefer spawning specialized agents for:

1. Complex multi-file changes requiring exploration
2. Tasks with clear, isolated scope (e.g., "write tests for module X")
3. Parallel work streams (feature A + bugfix B simultaneously)
4. Long-running analysis (codebase search, dependency audit)
5. Tasks requiring different contexts (multiple git worktrees)

Use the main conversation for:

1. User interaction and clarification
2. Quick single-file edits
3. Coordinating agent work
4. High-level decision making

Implementation

# Example: Spawn parallel test and documentation agents
task_tool(
    subagent_type="general-purpose",
    prompt="Write comprehensive tests for src/markets/schedule.py",
    description="Write schedule tests"
)

task_tool(
    subagent_type="general-purpose",
    prompt="Update README.md with global market feature documentation",
    description="Update README"
)

Use run_in_background=True for independent tasks that don't block subsequent work.

Common Command Failures

Critical: Learn from failures. Never repeat the same failed command without modification.

tea CLI (Gitea Command Line Tool)

❌ TTY Error - Interactive Confirmation Fails

~/bin/tea issues create --repo X --title "Y" --description "Z"
# Error: huh: could not open a new TTY: open /dev/tty: no such device or address

💡 Reason: tea tries to open /dev/tty for interactive confirmation prompts, which is unavailable in non-interactive environments.

✅ Solution: Use YES="" environment variable to bypass confirmation

YES="" ~/bin/tea issues create --repo jihoson/The-Ouroboros --title "Title" --description "Body"
YES="" ~/bin/tea issues edit <number> --repo jihoson/The-Ouroboros --description "Updated body"
YES="" ~/bin/tea pulls create --repo jihoson/The-Ouroboros --head feature-branch --base main --title "Title" --description "Body"

📝 Notes:

• Always set default login: ~/bin/tea login default local
• Use --repo jihoson/The-Ouroboros when outside repo directory
• tea is preferred over direct Gitea API calls for consistency

❌ Wrong Parameter Name

tea issues create --body "text"
# Error: flag provided but not defined: -body

💡 Reason: Parameter is --description, not --body.

✅ Solution: Use correct parameter name

YES="" ~/bin/tea issues create --description "text"

Gitea API (Direct HTTP Calls)

❌ Wrong Hostname

curl http://gitea.local:3000/api/v1/...
# Error: Could not resolve host: gitea.local

💡 Reason: Gitea instance runs on localhost:3000, not gitea.local.

✅ Solution: Use correct hostname (but prefer tea CLI)

curl http://localhost:3000/api/v1/repos/jihoson/The-Ouroboros/issues \
  -H "Authorization: token $GITEA_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"title":"...", "body":"..."}'

📝 Notes:

• Prefer tea CLI over direct API calls
• Only use curl for operations tea doesn't support

Git Commands

❌ User Not Configured

git commit -m "message"
# Error: Author identity unknown

💡 Reason: Git user.name and user.email not set.

✅ Solution: Configure git user

git config user.name "agentson"
git config user.email "agentson@localhost"

❌ Permission Denied on Push

git push origin branch
# Error: User permission denied for writing

💡 Reason: Repository access token lacks write permissions or user lacks repo write access.

✅ Solution:

1. Verify user has write access to repository (admin grants this)
2. Ensure git credential has correct token with write:repository scope
3. Check remote URL uses correct authentication

Python/Pytest

❌ Module Import Error

pytest tests/test_foo.py
# ModuleNotFoundError: No module named 'src'

💡 Reason: Package not installed in development mode.

✅ Solution: Install package with dev dependencies

pip install -e ".[dev]"

❌ Async Test Hangs

async def test_something():  # Hangs forever
    result = await async_function()

💡 Reason: Missing pytest-asyncio or wrong configuration.

✅ Solution: Already configured in pyproject.toml

[tool.pytest.ini_options]
asyncio_mode = "auto"

No decorator needed for async tests.

Build & Test Commands

# Install all dependencies (production + dev)
pip install ".[dev]"

# Run full test suite with coverage
pytest -v --cov=src --cov-report=term-missing

# Run a single test file
pytest tests/test_risk.py -v

# Run a single test by name
pytest tests/test_brain.py -k "test_parse_valid_json" -v

# Lint
ruff check src/ tests/

# Type check (strict mode, non-blocking in CI)
mypy src/ --strict

# Run the trading agent
python -m src.main --mode=paper

# Docker
docker compose up -d ouroboros          # Run agent
docker compose --profile test up test   # Run tests in container

Architecture

Self-evolving AI trading agent for Korean stock markets (KIS API). The main loop in src/main.py orchestrates five components in a 60-second cycle per stock:

1. Broker (src/broker/kis_api.py) — Async KIS API client with automatic OAuth token refresh, leaky-bucket rate limiter (10 RPS), and POST body hash-key signing. Uses a custom SSL context with disabled hostname verification for the VTS (virtual trading) endpoint due to a known certificate mismatch.

2. Brain (src/brain/gemini_client.py) — Sends structured prompts to Google Gemini, parses JSON responses into TradeDecision objects. Forces HOLD when confidence < threshold (default 80). Falls back to safe HOLD on any parse/API error.

3. Risk Manager (src/core/risk_manager.py) — READ-ONLY by policy (see docs/agents.md). Circuit breaker halts all trading via SystemExit when daily P&L drops below -3.0%. Fat-finger check rejects orders exceeding 30% of available cash.

4. Context Tree (src/context/) — NEW: Pillar 2 implementation. 7-tier hierarchical memory (L1-L7) from real-time quotes to generational wisdom. Auto-aggregates daily → weekly → monthly → quarterly → annual → legacy. See docs/context-tree.md for details.

5. Evolution (src/evolution/optimizer.py) — Analyzes high-confidence losing trades from SQLite, asks Gemini to generate new BaseStrategy subclasses, validates them by running the full pytest suite, and simulates PR creation.

Data flow per cycle: Fetch orderbook + balance → calculate P&L → query context tree → get Gemini decision → validate with risk manager → execute order → log to SQLite + context layers (src/db.py).

Key Constraints (from docs/agents.md)

• core/risk_manager.py is READ-ONLY. Changes require human approval.
• Circuit breaker threshold (-3.0%) may only be made stricter, never relaxed.
• Fat-finger protection (30% max order size) must always be enforced.
• Confidence < 80 must force HOLD — this rule cannot be weakened.
• All code changes require corresponding tests. Coverage must stay >= 80%.
• Generated strategies must pass the full test suite before activation.

Configuration

Pydantic Settings loaded from .env (see .env.example). Required vars: KIS_APP_KEY, KIS_APP_SECRET, KIS_ACCOUNT_NO (format XXXXXXXX-XX), GEMINI_API_KEY. Tests use in-memory SQLite (DB_PATH=":memory:") and dummy credentials via tests/conftest.py.

Test Structure

72 tests across five files. asyncio_mode = "auto" in pyproject.toml — async tests need no special decorator. The settings fixture in conftest.py provides safe defaults with test credentials and in-memory DB.

• test_risk.py (11) — Circuit breaker boundaries, fat-finger edge cases
• test_broker.py (6) — Token lifecycle, rate limiting, hash keys, network errors
• test_brain.py (18) — JSON parsing, confidence threshold, malformed responses, prompt construction
• test_market_schedule.py (19) — Market open/close logic, timezone handling, DST, lunch breaks
• test_context.py (18) — NEW: Context tree CRUD, aggregation logic, retention policies, layer metadata