agentleFS
MiaoDXMiaoDX/roboclaws/skills/raw-fpv-visual-labeler/SKILL.md

raw-fpv-visual-labeler

Skill7 starsChanged 53 days agoMITView on GitHub

Label visible cleanup-relevant movable objects from grouped RAW-FPV frames without creating executable cleanup handles.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill only for perception-only RAW-FPV labeling probes.

    line 8
  2. Distinctive:

    Use 3-6 neighboring RAW-FPV frames from the same waypoint, sweep segment, or source observation neighborhood when available.

    line 15
  3. Distinctive:

    Never include private labels, generated hidden target ids, acceptable destination truth, executable observed-object handles, detector …

    line 23

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

RAW-FPV Visual Labeler

Use this skill only for perception-only RAW-FPV labeling probes. It consumes public robot FPV frame evidence and emits structured visual labels for review and offline scoring. It does not call cleanup tools, does not create observed_* handles, and does not authorize navigation, pick, place, or done.

Inputs

Use 3-6 neighboring RAW-FPV frames from the same waypoint, sweep segment, or source observation neighborhood when available. Each frame may include only:

  • public frame id;
  • image artifact;
  • public waypoint or room context already visible to the cleanup agent;
  • optional public runtime-map planning hints marked non-executable.

Never include private labels, generated hidden target ids, acceptable destination truth, executable observed-object handles, detector candidates, or camera-label producer candidates.

Output

Return strict JSON:

{
  "schema": "raw_fpv_visual_labeler_response_v1",
  "labels": [
    {
      "evidence_frame_id": "run/raw_fpv_001",
      "category": "mug",
      "category_family": "dish",
      "coarse_region": "middle_right",
      "confidence": 0.82,
      "is_cleanup_relevant": true,
      "bbox": [0.62, 0.5, 0.12, 0.15],
      "surface_hint": "table",
      "reason_not_actionable": ""
    }
  ]
}

Required per label:

  • evidence_frame_id
  • category
  • category_family
  • coarse_region
  • confidence
  • is_cleanup_relevant

Optional per label:

  • bbox
  • surface_hint
  • reason_not_actionable

Allowed category families are food, dish, book, linen, toy, and electronics. Fixtures and surfaces such as tables, beds, counters, shelves, sinks, cabinets, and floors may be mentioned only as surface_hint or as is_cleanup_relevant=false; they are not object hits.

Boundary

These labels are perception evidence. The cleanup agent must not consume them as executable handles in camera-raw-fpv. A later assisted RAW-FPV or camera-grounded-labels producer decision would need its own contract change.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.