# operant-mcp MCP server

Read-only MCP server for the OPERANT AI operating-agent calibration benchmark.

## Links
- Registry page: https://www.getdrio.com/mcp/io-github-saagpatel-operant-mcp
- Repository: https://github.com/saagpatel/operant

## Install
- Command: `npx -y saagar-operant-mcp`
- Endpoint: https://operant-mcp.saagar210.workers.dev/mcp
- Auth: Not captured

## Setup notes
- Package: Npm saagar-operant-mcp v0.1.0
- Remote endpoint: https://operant-mcp.saagar210.workers.dev/mcp

## Tools
- get_results (Get retained calculation profiles) - Return retained model calculation profiles plus generated_at, claim_status, claims_at_risk, and the bounded evidence_binding. These rows are not durable named-model performance claims. Do not rank them or treat stdev/significance as reliable model evidence. Endpoint: https://operant-mcp.saagar210.workers.dev/mcp
- compare_models (Inspect two retained profiles) - Place two retained calculation profiles side by side by display_name substring. Returns ocs_mean, ocs_stdev, orchestration_mean, run_family, and subject_shell for each, plus comparison_status=NOT_DURABLE and the public claim_status. This is not evidence that either named model outperforms, equals, or differs significantly from the other. If a name is ambiguous or not found, returns an error listing all available display_names. Endpoint: https://operant-mcp.saagar210.workers.dev/mcp
- get_methodology (Get benchmark methodology) - Return the benchmark design: the 4 axes, the OCS formula (Youden's J), the 5 decision labels (PROCEED, PROCEED_SANCTIONED, REFUSE, ESCALATE, REROUTE), scoring blocks, and a concise what-it-measures / what-it-doesn't summary. Sourced from the baked case-file metadata. Endpoint: https://operant-mcp.saagar210.workers.dev/mcp
- list_cases (List cases) - Return case metadata (no full task prompts): pair_id/id, axis, tier, grounding, and side indicators (malign/benign for axes 1/2/4; null for axis 3). Filter by axis, or omit for all cases across all axes (the result includes a count). Use get_case to fetch a full case with task prompts and expected decisions. Endpoint: https://operant-mcp.saagar210.workers.dev/mcp
- get_case (Get a case) - Return the full case for a given pair_id (axes 1/2/4) or id (axis 3): malign and benign task prompts, expected decisions, grounding rationale, and bypass patterns. Axis 3 cases are single (unmatched) and use an 'id' field instead of 'pair_id'. Use list_cases to browse available ids. Endpoint: https://operant-mcp.saagar210.workers.dev/mcp

## Resources
- operant://results - Retained Calculation Profiles Historical/local calculation profiles with the public integrity boundary (JSON) MIME type: application/json
- operant://methodology - Methodology OPERANT benchmark design, axes, decision labels, and OCS formula (JSON) MIME type: application/json

## Prompts
- score_my_agent - Score my agent with OPERANT A ready prompt explaining how to run OPERANT against your own Claude Code agent and how to read the OCS score.

## Metadata
- Owner: io.github.saagpatel
- Version: 0.1.0
- Runtime: Npm
- Transports: STDIO, HTTP
- License: Not captured
- Language: Not captured
- Stars: Not captured
- Updated: Jun 28, 2026
- Source: https://registry.modelcontextprotocol.io