# active_learning.plan_retraining

Create a correction dataset and offline training run against a fixed benchmark.

Agent View of the PolicyLayer registry record for `active_learning.plan_retraining`. HTML page: https://policylayer.com/tools/visionmcp/active-learning.plan-retraining

## Facts

- Tool: `active_learning.plan_retraining`
- Server: Visionmcp (`joshuahickscorp/visionmcp`) — https://policylayer.com/tools/visionmcp.md
- Homepage: https://github.com/joshuahickscorp/visionmcp
- Risk category: Execute (High 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": "active_learning.plan_retraining",
    "arguments": {}
  }
}
```

## Why active_learning.plan_retraining is rated High

This tool triggers an external operation (model retraining) whose effects are significant and depend on arguments (the correction dataset, benchmark, and training parameters). While it doesn't delete data (not Destructive) or move money (not Financial), it executes a complex computational process that modifies ML model state.

From the tool's own definition: "Tool performs 'offline training run' which executes machine learning operations that modify model state and behavior."

## Use case

AI agents invoke active_learning.plan_retraining to trigger actions in Visionmcp. What it does depends on the arguments the agent supplies, and its effects often reach beyond the immediate call: builds kicked off, notifications sent, workflows started.

## Recommended policy (PolicyLayer)

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

```json
{
  "version": "1",
  "default": "deny",
  "tools": {
    "active_learning.plan_retraining": {
      "limits": [
        {
          "counter": "active_learning.plan_retraining_rate",
          "window": "minute",
          "max": 10,
          "scope": "grant"
        }
      ]
    }
  }
}
```

## Other tools on Visionmcp (375)

- `active_learning.rollback` — Destructive — https://policylayer.com/tools/visionmcp/active-learning.rollback.md
- `artifact.reap_stale_uploads` — Destructive — https://policylayer.com/tools/visionmcp/artifact.reap-stale-uploads.md
- `active_learning.execute_retraining` — Execute — https://policylayer.com/tools/visionmcp/active-learning.execute-retraining.md
- `benchmark.bootstrap_calibration` — Execute — https://policylayer.com/tools/visionmcp/benchmark.bootstrap-calibration.md
- `benchmark.bootstrap_dgx_spark` — Execute — https://policylayer.com/tools/visionmcp/benchmark.bootstrap-dgx-spark.md
- `benchmark.bootstrap_rtx_5090_fe` — Execute — https://policylayer.com/tools/visionmcp/benchmark.bootstrap-rtx-5090-fe.md
- `benchmark.refine_dgx_spark_base_foot_candidate` — Execute — https://policylayer.com/tools/visionmcp/benchmark.refine-dgx-spark-base-foot-candidate.md
- `benchmark.refine_dgx_spark_visual_candidate` — Execute — https://policylayer.com/tools/visionmcp/benchmark.refine-dgx-spark-visual-candidate.md
- `benchmark.refine_rtx_5090_fe_front_frame_candidate` — Execute — https://policylayer.com/tools/visionmcp/benchmark.refine-rtx-5090-fe-front-frame-candidate.md
- `benchmark.refine_rtx_5090_fe_visual_candidate` — Execute — https://policylayer.com/tools/visionmcp/benchmark.refine-rtx-5090-fe-visual-candidate.md
- `blender.generate_lod` — Execute — https://policylayer.com/tools/visionmcp/blender.generate-lod.md
- `blender.prepare_asset` — Execute — https://policylayer.com/tools/visionmcp/blender.prepare-asset.md
- `blender.render` — Execute — https://policylayer.com/tools/visionmcp/blender.render.md
- `camera.derive_undistorted` — Execute — https://policylayer.com/tools/visionmcp/camera.derive-undistorted.md
- `campaign.advance` — Execute — https://policylayer.com/tools/visionmcp/campaign.advance.md
- `campaign.resume` — Execute — https://policylayer.com/tools/visionmcp/campaign.resume.md
- `campaign.start` — Execute — https://policylayer.com/tools/visionmcp/campaign.start.md
- `candidate.evaluate_transaction` — Execute — https://policylayer.com/tools/visionmcp/candidate.evaluate-transaction.md
- `component.generate` — Execute — https://policylayer.com/tools/visionmcp/component.generate.md
- `coverage.acquire_missing` — Execute — https://policylayer.com/tools/visionmcp/coverage.acquire-missing.md
- `dataset.generate` — Execute — https://policylayer.com/tools/visionmcp/dataset.generate.md
- `dataset.train_feature_model` — Execute — https://policylayer.com/tools/visionmcp/dataset.train-feature-model.md
- `evidence.discover` — Execute — https://policylayer.com/tools/visionmcp/evidence.discover.md
- `geometry.run_ensemble` — Execute — https://policylayer.com/tools/visionmcp/geometry.run-ensemble.md
- `photogrammetry.run` — Execute — https://policylayer.com/tools/visionmcp/photogrammetry.run.md
- `portfolio.execute_initial` — Execute — https://policylayer.com/tools/visionmcp/portfolio.execute-initial.md
- `portfolio.execute_parametric_seed` — Execute — https://policylayer.com/tools/visionmcp/portfolio.execute-parametric-seed.md
- `recon.bootstrap` — Execute — https://policylayer.com/tools/visionmcp/recon.bootstrap.md
- `recon.continue_multiview_search` — Execute — https://policylayer.com/tools/visionmcp/recon.continue-multiview-search.md
- `reference.propose_masks` — Execute — https://policylayer.com/tools/visionmcp/reference.propose-masks.md
- …and 345 more: https://policylayer.com/tools/visionmcp.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=visionmcp · API: https://policylayer.com/registry/api · Policy library: https://policylayer.com/policies/visionmcp