CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

train-pose

Train/evaluate WiFi pose models honestly — camera-supervised (MediaPipe + CSI) and camera-free (WiFlow), always checked against the mean-pose baseline before any PCK is quoted.

Quality

82%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

SKILL.md
Quality
Evals
Security

Quality

Content

86%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A tight, high-signal body: honest-reporting discipline is front-loaded, concrete harness commands are copy-paste ready, and the publish checklist provides explicit validation. The only meaningful gaps are the missing command for running the mean-pose baseline itself and the absence of an explicit fix-and-retry loop when the gate fails.

DimensionReasoningScore

Conciseness

The body is lean and assumes competence: 'Build a CSI→pose model without overstating it' opens directly, no concepts are re-explained, and each section (baseline, paths, harness commands, publish checklist) carries only non-obvious project-specific facts. Every token earns its place; not 4 because there is no over-explanation to trim.

5 / 5

Actionability

The harness section gives four copy-paste-ready commands with flags ('npx @ruvnet/ruview train --mode pose --data-dir <in-repo MM-Fi dir> --confirm') and the publish checklist is concrete. It is not fully at the 5 anchor: the mean-pose baseline run itself has no shown command, and 'ruview_claim_check the writeup' names a tool without an invocation example, leaving minor execution gaps.

4 / 5

Workflow Clarity

A clear sequence is present (plan → smoke train → real train → gate → publish checklist) with explicit validation checkpoints ('train-gate --file eval-report.json', the claim check, 'Quote nothing it fails'). It falls short of 5 because there is no error-recovery loop: what to do when the gate or claim check fails is only implied (suppress the claim), not an explicit fix-and-retry path.

4 / 5

Progressive Disclosure

The body is ~42 lines, single-purpose (honest pose-model training/evaluation), and no bundle files exist or are needed; per the under-50-lines exception, well-organized sections alone justify the top score. Its headers (baseline, Paths, harness, publish checklist) make navigation trivial and the ADR tags are inline, not dangling file references.

5 / 5

Total

18

/

20

Passed

Description

70%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A specific, distinctive description that clearly states what the skill does across its two model paths and its honesty discipline, using strong domain trigger terms. Its main weakness is the absence of any 'Use when...' trigger clause, which both caps completeness and leaves invocation guidance implicit.

Suggestions

Add an explicit trigger clause, e.g., 'Use when training or evaluating WiFi/CSI pose models, quoting PCK numbers, or comparing against camera-supervised baselines.'

Include a few more natural synonyms users might say (e.g., 'human pose estimation', 'wireless sensing', 'keypoint accuracy') to broaden trigger coverage.

State the delivery context (e.g., that results are produced via the ruview harness commands) so the 'what' covers how the skill executes, not just what it produces.

DimensionReasoningScore

Specificity

The description lists several concrete actions — 'Train/evaluate WiFi pose models', 'checked against the mean-pose baseline before any PCK is quoted' — and names the two concrete paths (camera-supervised with MediaPipe + CSI, camera-free with WiFlow). It falls short of the comprehensive anchor because supporting actions like reporting/gating the claim are only implied, and short of no gaps, but it goes well beyond the 1-2-action anchor.

4 / 5

Completeness

It clearly answers 'what' (train/evaluate WiFi pose models across camera-supervised and camera-free paths, always baseline-checked) but contains no 'Use when...' clause or equivalent explicit trigger guidance; the 'when' is at best weakly implied by the domain phrasing. Per the judging guideline, a missing 'Use when' clause caps completeness at 3.

3 / 5

Trigger Term Quality

Natural terms a user in this domain would say are present: 'WiFi pose', 'pose models', 'train', 'evaluate', 'MediaPipe', 'CSI', 'WiFlow', 'PCK'. A few natural variations are missing (e.g., 'human pose estimation', 'wireless sensing', 'keypoints'), so it is good-but-not-comprehensive keyword coverage.

4 / 5

Distinctiveness Conflict Risk

The description occupies a clear niche — WiFi CSI-based pose estimation with named, distinctive anchors (WiFlow, MediaPipe + CSI pairing, PCK gating against a mean-pose baseline) — making it highly distinguishable from generic ML or vision skills with minimal conflict risk.

5 / 5

Total

16

/

20

Passed

Validation

100%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository

ruvnet/RuView

Path

harness/ruview/.claude/skills/train-pose/SKILL.md

Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.