trader-mcp

Trader MCP

Trader MCP runs strategies, persists their state, and exposes a compact MCP surface for operator and Hermes control.

Current shape

• strategies are bound to an account and market
• strategy snapshots expose report() for Hermes
• Hermes policy is stored separately and mapped into strategy-specific parameters
• market regime is read from crypto-mcp, not Trader-owned data

What works now

• strategy lifecycle control (start, pause, resume, stop, reconcile)
• strategy snapshots with report() and fit
• Hermes policy storage via set_strategy_policy
• Hermes action dispatch via apply_control_decision(payload)
• grid trading
• exposure protection
• trend following
• dashboard inspection and inline config editing

Public MCP tools

• list_strategies
• get_strategy
• update_strategy
• control_strategy
• set_strategy_policy
• apply_control_decision
• get_capabilities

get_strategy() defaults

By default, get_strategy() returns compact identity/control data plus report. Optional flags:

• include_config
• include_state
• include_render
• include_debug
• include_report (defaults to true)

Quick start

source .venv/bin/activate
pip install -r requirements.txt
./run.sh

Default port: 8570

Endpoints

• GET / landing page
• GET /health health check
• GET /mcp/sse MCP SSE transport endpoint
• GET /dashboard/ dashboard

Notes

• control_strategy() handles lifecycle verbs.
• set_strategy_policy() stores Hermes intent as risk_posture and priority.
• apply_control_decision() is the canonical Hermes write tool for switch, pause, resume, and set_risk_mode.
• report().fit is the main Hermes-facing strategy fit block.
• report().supervision carries the strategy self-report used by Hermes for switching safety and ranking.
• policy is applied again on reconcile.

Development

See run.sh, tests.sh, killserver.sh, and restart.sh.