# calibration.optimize_rubrics

Tune evaluation rubrics for better alignment with human judgments

Agent View of the PolicyLayer registry record for `calibration.optimize_rubrics`. HTML page: https://policylayer.com/tools/jrmatherly-mcp-context-forge/calibration.optimize-rubrics

## Facts

- Tool: `calibration.optimize_rubrics`
- Server: ContextForge MCP Gateway (`jrmatherly/mcp-context-forge`) — https://policylayer.com/tools/jrmatherly-mcp-context-forge.md
- Homepage: https://github.com/jrmatherly/mcp-context-forge
- Risk category: Write (Medium risk)
- Registry record: grade F, identity unverified
- Server rate-limited: no
- Parameters: 0
- Recommended policy verdict: Rate-limited

## Example call (MCP tools/call, JSON-RPC 2.0)

```json
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "calibration.optimize_rubrics",
    "arguments": {}
  }
}
```

## Why calibration.optimize_rubrics is rated Medium

This tool modifies evaluation rubrics to optimize them for alignment with human judgments. It is a Write operation as it updates/tunes configuration or scoring criteria. Misuse could cause AI evaluation pipelines to be miscalibrated, potentially skewing quality assessments, but it does not execute code, delete data, or involve finances.

From the tool's own definition: "'Tune evaluation rubrics' and 'better alignment with human judgments' indicate modification of rubric/evaluation configuration"

## Use case

AI agents use calibration.optimize_rubrics to create or update resources in ContextForge MCP Gateway, usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your ContextForge MCP Gateway environment.

## Recommended policy (PolicyLayer)

Verdict: **Rate-limited**. Enforced by the PolicyLayer MCP gateway (https://policylayer.com/mcp-gateway) before a call reaches ContextForge MCP Gateway:

```json
{
  "version": "1",
  "default": "deny",
  "tools": {
    "calibration.optimize_rubrics": {
      "limits": [
        {
          "counter": "calibration.optimize_rubrics_rate",
          "window": "minute",
          "max": 30,
          "scope": "grant"
        }
      ]
    }
  }
}
```

## Other tools on ContextForge MCP Gateway (220)

- `clear_cache` — Destructive — https://policylayer.com/tools/jrmatherly-mcp-context-forge/clear-cache.md
- `delete_slide` — Destructive — https://policylayer.com/tools/jrmatherly-mcp-context-forge/delete-slide.md
- `agent.benchmark_performance` — Execute — https://policylayer.com/tools/jrmatherly-mcp-context-forge/agent.benchmark-performance.md
- `bitbucket_trigger_pipeline` — Execute — https://policylayer.com/tools/jrmatherly-mcp-context-forge/bitbucket-trigger-pipeline.md
- `build_schedule` — Execute — https://policylayer.com/tools/jrmatherly-mcp-context-forge/build-schedule.md
- `calculate_stats` — Execute — https://policylayer.com/tools/jrmatherly-mcp-context-forge/calculate-stats.md
- `compile_document` — Execute — https://policylayer.com/tools/jrmatherly-mcp-context-forge/compile-document.md
- `critical_path_analysis` — Execute — https://policylayer.com/tools/jrmatherly-mcp-context-forge/critical-path-analysis.md
- `execute_code` — Execute — https://policylayer.com/tools/jrmatherly-mcp-context-forge/execute-code.md
- `generate_qr_code` — Execute — https://policylayer.com/tools/jrmatherly-mcp-context-forge/generate-qr-code.md
- `judge.rank_responses` — Execute — https://policylayer.com/tools/jrmatherly-mcp-context-forge/judge.rank-responses.md
- `performance.evaluate_throughput_scaling` — Execute — https://policylayer.com/tools/jrmatherly-mcp-context-forge/performance.evaluate-throughput-scaling.md
- `query_data` — Execute — https://policylayer.com/tools/jrmatherly-mcp-context-forge/query-data.md
- `render_graph` — Execute — https://policylayer.com/tools/jrmatherly-mcp-context-forge/render-graph.md
- `robustness.assess_distribution_shift` — Execute — https://policylayer.com/tools/jrmatherly-mcp-context-forge/robustness.assess-distribution-shift.md
- `robustness.measure_input_sensitivity` — Execute — https://policylayer.com/tools/jrmatherly-mcp-context-forge/robustness.measure-input-sensitivity.md
- `robustness.test_adversarial_inputs` — Execute — https://policylayer.com/tools/jrmatherly-mcp-context-forge/robustness.test-adversarial-inputs.md
- `statistical_test` — Execute — https://policylayer.com/tools/jrmatherly-mcp-context-forge/statistical-test.md
- `transform_data` — Execute — https://policylayer.com/tools/jrmatherly-mcp-context-forge/transform-data.md
- `workflow.run_evaluation` — Execute — https://policylayer.com/tools/jrmatherly-mcp-context-forge/workflow.run-evaluation.md
- `agent.analyze_reasoning` — Read — https://policylayer.com/tools/jrmatherly-mcp-context-forge/agent.analyze-reasoning.md
- `agent.evaluate_tool_use` — Read — https://policylayer.com/tools/jrmatherly-mcp-context-forge/agent.evaluate-tool-use.md
- `agent.measure_task_completion` — Read — https://policylayer.com/tools/jrmatherly-mcp-context-forge/agent.measure-task-completion.md
- `analyze_code` — Read — https://policylayer.com/tools/jrmatherly-mcp-context-forge/analyze-code.md
- `analyze_csv` — Read — https://policylayer.com/tools/jrmatherly-mcp-context-forge/analyze-csv.md
- `analyze_dataset` — Read — https://policylayer.com/tools/jrmatherly-mcp-context-forge/analyze-dataset.md
- `analyze_document` — Read — https://policylayer.com/tools/jrmatherly-mcp-context-forge/analyze-document.md
- `analyze_feed` — Read — https://policylayer.com/tools/jrmatherly-mcp-context-forge/analyze-feed.md
- `analyze_graph` — Read — https://policylayer.com/tools/jrmatherly-mcp-context-forge/analyze-graph.md
- `analyze_text` — Read — https://policylayer.com/tools/jrmatherly-mcp-context-forge/analyze-text.md
- …and 190 more: https://policylayer.com/tools/jrmatherly-mcp-context-forge.md

## For agents

This record is a snapshot. Live verdicts and the full registry:

- Check every server in your MCP config at once: `npx -y policylayer stack`
- Vet a server before you add it: install the mcp-precheck skill — `npx skills add https://policylayer.com` (skill text: https://policylayer.com/skill.md)
- Query the registry over MCP: endpoint `https://api.policylayer.com/mcp` — tools `check_mcp_server`, `check_mcp_stack`, `check_tool`, `search_registry`, `get_change_events`

---

Source: the PolicyLayer MCP registry — one continuously verified record per MCP server. Full record: https://policylayer.com/registry?q=jrmatherly-mcp-context-forge · API: https://policylayer.com/registry/api · Policy library: https://policylayer.com/policies/jrmatherly-mcp-context-forge