Tools reference — riskmanaged.io docs

The full list of the 70+ MCP tools your AI assistant can call, grouped by category, plus the signal-condition JSON shape.

The MCP server exposes 70+ tools organized by category. Your AI assistant can call any of these tools. Use this page when you want to know whether a particular platform operation is reachable from your client.

Strategy management#

Tool	Description
`list_strategies`	List strategies with optional filters (mode, ticker, timeframe)
`get_strategy`	Get full strategy configuration by ID
`create_strategy`	Create a new strategy with name, ticker, timeframe, exchange
`update_strategy`	Update strategy properties
`delete_strategy`	Delete a strategy

Indicators#

Tool	Description
`list_indicator_types`	List all available indicator types
`get_indicator_schema`	Get the JSON schema for an indicator type
`add_indicator`	Add an indicator to a strategy
`update_indicator`	Update indicator parameters
`remove_indicator`	Remove an indicator from a strategy

Signals & bias#

Tool	Description
`add_signal_group`	Create a new signal group
`add_signal_rule`	Add a rule to a signal group
`edit_signal_rule`	Modify an existing rule
`delete_signal_rule`	Remove a rule
`add_bias_generator`	Create a bias generator
`add_bias_rule`	Add a bias rule

Signal rule conditions#

add_signal_rule, edit_signal_rule and add_bias_rule take a conditions array — the JSON form of what the builder's Condition rows express (see signals for the UI version). Each leaf compares one line against a constant or another line:

{
  "trigger_action": "enter_position",
  "trigger_direction": "long",
  "conditions": [
    { "and": [
      { "trigger_line": "RSI.rsi", "trigger": "lt", "threshold_value": 30 }
    ] }
  ]
}

Field	Meaning
`trigger_line`	The line being tested, e.g. `RSI.rsi`, or a price line (`close`)
`trigger`	Operator: `crossover`, `crossunder`, `gt`, `ge`, `lt`, `le`, `eq`
`threshold_value`	Compare against a constant — or —
`threshold_line`	compare against another line, e.g. `MACD.signal`
`trigger_line_shift` / `threshold_line_shift`	Optional look-back in candles (default `0`)

crossover / crossunder are the event operators (fire once on the cross); gt / lt and friends are state operators (true while the condition holds). A bias rule uses the same shape but takes a bias_direction (long / short / neutral) instead of trigger_action.

Risk management#

Tool	Description
`set_stop_loss`	Configure stop-loss (simple, trailing, or ATR)
`set_take_profit`	Configure take-profit (spread)
`remove_stop_loss`	Remove stop-loss configuration
`remove_take_profit`	Remove take-profit configuration

Backtesting#

Tool	Description
`run_backtest`	Run a backtest for a strategy
`get_backtest_results`	Get backtest metrics

Grid search#

The grid tools model the full research loop: create a template, check the variation count, launch it, wait for it to finish, analyze the robustness verdict, then refine or label the winner. See the "Research a strategy end-to-end" workflow in MCP workflows.

Tool	Description
`create_grid_template`	Create a template from a strategy. `kind` picks the range shape and every mode is auto-clamped to the plan cap: `zoom_in` (default — narrow around current values), `explore_higher` / `explore_lower` (full window above / below), `manual` (legacy wide default).
`list_grid_templates`	List every grid template the caller owns
`update_grid_template`	Modify a template's parameter ranges before launch
`delete_grid_template`	Delete a template
`check_variations`	Return `{count, limit}` for a template — always call before `create_grid`
`create_grid`	Launch the grid; runs every variation, costs tokens beyond the daily allowance
`list_grids`	The caller's grids with filters. `label`, `classification`, `root_ticker/timeframe/exchange`, `state`, `search`, plus numeric filters. Two numeric shapes: operator suffixes `<col>_<op>` where col ∈ {`sharpe`, `cumulative_return`, `max_drawdown`, `time_in_market`, `robustness_score`} and op ∈ {`gt`, `gte`, `lt`, `lte`} — e.g. `sharpe_gte=1.5`, `max_drawdown_gte=-0.15`. Legacy aliases `min_sharpe`, `max_sharpe`, `min_cumulative_return`, `max_cumulative_return`, `min_robustness_score`, and `max_drawdown` (magnitude — `0.15` = "no worse than -15%") still work. Sort by `robustness / sharpe / cumulative_return / max_drawdown / time_in_market`.
`get_grid`	Grid detail: robustness verdict, ranked clusters (each with width + members), every variation's metrics
`analyze_grid`	(Re)compute the cluster analysis — used to backfill older grids; new ones analyse automatically
`get_grid_suggestions`	Proposed next searches from a finished grid — `zoom_in` inside the plateau + `explore_higher`/`_lower` above/below the tested range
`refine_grid`	Persist a chosen next-search as a new template; returns `{template_id}`. Follow with `create_grid` to launch
`archive_grid`	Hide a grid from the default list; reversible in the UI
`set_grid_labels`	Replace a grid's labels (free-form user-scoped tags, max 20 per grid, 32 chars each). Pass the full desired list; the call replaces rather than merges.
`list_grid_labels`	Autocomplete — every distinct label the caller has ever used

Versioning & community#

Tool	Description
`commit_version`	Commit a version snapshot
`list_versions`	List version history
`restore_version`	Restore a previous version
`share_strategy`	Share strategy to community

Hedge-fund templates#

Tool	Description
`list_templates`	List the 3 day-1 templates
`get_template`	Get one template's full cast + defaults

Agent committees#

Tool	Description
`list_committees`	List user's Squads
`get_committee`	Get one Squad (with members, covered markets and readings)
`clone_template`	Clone a day-1 template into a working Squad
`trigger_committee_run`	Run one deliberation on one market
`get_committee_messages`	Replay the deliberation message bus

Model routes#

Tool	Description
`list_model_routes`	List user's per-task-class LLM routes
`upsert_model_route`	Create or update a route
`delete_model_route`	Delete a route

LLM connections#

Tool	Description
`list_llm_connections`	List user's LLM credentials (platform + direct + remote)
`create_llm_connection`	Create a direct (user-provided) LLM connection
`test_llm_connection`	Ping a connection to verify the credentials
`reveal_llm_connection_key`	Reveal the decrypted API key (use sparingly)
`update_llm_connection`	Update a connection (label, endpoint, is_active)
`delete_llm_connection`	Delete a direct or remote LLM connection

News#

Tool	Description
`list_news_articles`	List recent news articles (filter by ticker, source, date)
`get_news_article`	Get one news article by id
`list_news_sources`	List user's news sources (RSS, CryptoPanic, Fear&Greed)

Macro#

Tool	Description
`list_macro_events`	List macro events (FOMC, CPI, etc.) with date filter
`get_macro_event`	Get one macro event by id

User settings & daily spend cap#

Tool	Description
`get_user_settings`	Get the user's daily token cap + today's spend so far
`set_daily_token_cap`	Set or clear the daily token cap (null to disable)

Committee cadence and markets#

Tool	Description
`set_committee_cadence`	How often it wakes (30-minute floor), the scoring deadband, and how many bars of each reading the specialists see
`list_committee_markets`	The markets it covers — one decision each per wake
`add_committee_market`	Cover a market. Idempotent — re-adding re-enables a paused one
`remove_committee_market`	Stop covering a market. Past decisions are kept
`get_committee_readings`	The indicator readings fed to the specialists
`set_committee_readings`	Replace the whole reading map

Committee decisions#

Tool	Description
`get_committee_decision_board`	One row per covered market: current call, points, next wake
`list_committee_decisions`	Recent decisions, newest first. The newest is always unscored
`get_committee_decision_summary`	Standing: cumulative points, recent form, accuracy, streak

NOTE

There are no proposal or promotion tools. Squads do not raise trades for approval and have no autonomy tier — they publish a decision per covered market, and strategies read it through the Committee Decision indicator. See using a Squad in a strategy.