tradememory — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited tradememory (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Two things for an agent that trades:
how they turned out, and get told when you are tilting: losing streaks, oversized positions after a loss, drawdown past your own line. Every record is SHA-256 chained and anchored daily to an RFC 3161 timestamp authority, so a changed record is detectable without trusting our clock.
MCP server. Orders are evaluated against your policy before they reach the broker; refusals, approvals and replays are all recorded. Alpaca is the first broker. See "Brake" below.
Local-first: SQLite on your machine, no telemetry, no hosted account needed. MIT licensed, 1,500+ tests, CI on Python 3.10 to 3.12.
pip install tradememory-protocol
# brake preview (Python 3.12+):
pip install "tradememory-protocol[proxy]"Verify: tradememory doctor.
{
"mcpServers": {
"tradememory": { "command": "uvx", "args": ["tradememory-protocol"] }
}
}claude mcp add tradememory -- uvx tradememory-protocolUse this skill; the server command is uvx tradememory-protocol (stdio).
tradememory proxy init --account-id <alpaca account id> --symbols AAPL,MSFT
tradememory proxy doctor --env-file ~/.secrets/alpaca-paper.env # checks tool names and the account id
tradememory proxy config # the MCP entry that replaces the direct Alpaca oneThen point the agent at tradememory proxy run --policy ... --env-file ... instead of uvx alpaca-mcp-server. The agent sees all of Alpaca's tools plus recall_memories, get_behavioral_analysis, get_agent_state, brake_status.
Refused by default: symbols outside your list, orders above your notional or position limits, entries without a bracket stop, new orders after your daily loss or drawdown limit, and everything after tradememory proxy halt FULL_HALT. Never blocked: closing a position. Orders at or above approval_notional wait for tradememory proxy approve <intent_id>; the agent retries with the same client_order_id and the proxy forwards it exactly once. Anything the brake cannot evaluate is refused, not passed through. Full walkthrough with real outputs: docs/recipes/alpaca-brake.md.
| Group | Tools |
|---|---|
| Memory | remember_trade, recall_memories, get_trade_reflection, get_strategy_performance |
| State and behaviour | get_agent_state, get_behavioral_analysis, check_trade_legitimacy, compute_dqs |
| Plans | create_trading_plan, check_active_plans |
| Audit | export_audit_trail, verify_audit_hash, verify_audit_chain, get_daily_root |
| Validation | validate_strategy (deflated Sharpe, walk-forward, regime, CPCV) |
| Evolution (research-stage) | evolution_fetch_market_data, evolution_discover_patterns, evolution_run_backtest, evolution_evolve_strategy, evolution_get_log |
The Evolution Engine and the DQS gate are research-stage: see LIMITATIONS.md for what has and has not survived validation before you trust either for a trading decision.
"Record my AAPL long at 195, earnings beat, high confidence."
"Before I buy AAPL again: what happened last time in this condition?"
"Am I on tilt? Check my state and my sizing after losses."
"Verify the audit chain."
"What does the brake refuse right now?" (brake_status, proxy only)except the daily RFC 3161 timestamp request (configurable) and, if you set ANTHROPIC_API_KEY, reflection calls to the Anthropic API. The Evolution Engine fetches OHLCV from Binance's public API when you use it.
to the broker process it starts (uvx alpaca-mcp-server). It never writes keys to disk or to the database.
TRADEMEMORY_DB), a policy file and a smallstate file under ~/.tradememory/.
no elevation.
| Skill | Path | Description |
|---|---|---|
| Strategy Validator | .skills/strategy-validator/SKILL.md | Validate a backtest for overfitting with four statistical tests. Use when the user says "validate my strategy" or "is this overfitting?". |
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.