multiview-fit-loop Eval
=======================

Status: review
Score: 65/100
Risk: medium
Decision: manual_review
Policy: review
Reason: Test manually in an isolated workspace and compare against safer alternatives.

Install:
npx skills add CheshireJCat/blender --skill multiview-fit-loop

Required checks:
- PASS Task fit: Task wording matches this skill metadata.
- PASS Install path: Install handoff is available.
- PASS Install command safety: standard package or runtime install path
- WARN Trust score: Potentially useful, but at least one trust signal needs human inspection.
- WARN Audit score: Needs review
- WARN Agent safety gate: Sparse or mixed signals. Useful for discovery, but not for autonomous installation.
- PASS License clarity: MIT
- WARN Permission surface: shell or command execution, filesystem or document access

Warnings:
- Trust score: Potentially useful, but at least one trust signal needs human inspection.
- Audit score: Needs review
- Agent safety gate: Sparse or mixed signals. Useful for discovery, but not for autonomous installation.
- Permission surface: shell or command execution, filesystem or document access
- High-risk permission hints: Shell or command execution
- SKILL.md lacks setup/installation instructions and a concrete CLI example for running scripts/multiview_fit_report.py; dependencies such as OpenCV and NumPy are not mentioned.
- The hard gates in SKILL.md specify stricter front-view bbox size tolerance (3%) versus side/top/back (5%), but the script only implements a uniform 5% pass gate.
- The script's summary can report all views passing even when only one view was supplied, because it does not enforce that every required view is present.
- Allowed-tools lists blender_scene_info twice, which is a minor metadata inconsistency.
- The constraint inconsistency gate is described in SKILL.md but is not automated or checked by the provided script; it remains a manual process.
- Low GitHub adoption signal
- Quality score needs review

Validation plan:
1. Inspect repository, README/SKILL.md, license, and recent commits before production use.
2. Install in an isolated workspace or sandbox with no production secrets available.
3. Run the smallest representative task and record files touched, commands run, network access, and outputs.
4. Compare the selected skill against at least one alternative when the eval status is review or failed.
5. Promote only after the agent reports a successful verification result and unresolved warnings are accepted.

Do not use when:
- teams that need a vendor-supported SLA
- production agents without a repository review
- Low GitHub adoption signal
- SKILL.md lacks setup/installation instructions and a concrete CLI example for running scripts/multiview_fit_report.py; dependencies such as OpenCV and NumPy are not mentioned.
- No OpenAgentSkill engagement data yet
- High-risk permission hints: Shell or command execution
- The hard gates in SKILL.md specify stricter front-view bbox size tolerance (3%) versus side/top/back (5%), but the script only implements a uniform 5% pass gate.
- The script's summary can report all views passing even when only one view was supplied, because it does not enforce that every required view is present.

URLs:
- Skill: https://www.openagentskill.com/skills/cheshirejcat-multiview-fit-loop
- Audit: https://www.openagentskill.com/skills/cheshirejcat-multiview-fit-loop/audit
- JSON: https://www.openagentskill.com/api/agent/evals?slug=cheshirejcat-multiview-fit-loop