autoresearch Eval
=================

Status: failed
Score: 65/100
Risk: high
Decision: do_not_auto_install
Policy: block
Reason: Install path: No install command or repository handoff is available.

Install:


Required checks:
- PASS Task fit: Task wording matches this skill metadata.
- FAIL Install path: No install command or repository 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.
- FAIL Audit score: Risky
- FAIL Agent safety gate: This skill should not be selected by an agent without explicit human security review.
- 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.
- Permission surface: shell or command execution, filesystem or document access
- Audit risk risky exceeds max_risk=medium
- High-risk permission hints: Shell or command execution
- Potential broker, wallet, exchange, or real-money execution surface; sandbox and explicit approval are required
- The 'NEVER STOP' directive could be risky if the user forgets to interrupt, but the skill includes a max_rounds setting (default 50) to bound the loop, mitigating this concern.
- The skill relies on subprocess isolation and git revert, which are good practices, but the documentation could be clearer on how to handle unexpected errors during the loop.
- Low GitHub adoption signal
- This skill may touch real-money trading, broker, wallet, or exchange operations; use only in a sandbox with explicit approval.
- Quality score needs review
- GitHub adoption: 42 GitHub stars
- Stars/forks activity: 42 stars, 3 forks; issue activity unavailable in current metadata

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
- The 'NEVER STOP' directive could be risky if the user forgets to interrupt, but the skill includes a max_rounds setting (default 50) to bound the loop, mitigating this concern.
- No OpenAgentSkill engagement data yet
- Audit risk risky exceeds max_risk=medium
- High-risk permission hints: Shell or command execution
- Potential broker, wallet, exchange, or real-money execution surface; sandbox and explicit approval are required

URLs:
- Skill: https://www.openagentskill.com/skills/darkroomengineering-autoresearch
- Audit: https://www.openagentskill.com/skills/darkroomengineering-autoresearch/audit
- JSON: https://www.openagentskill.com/api/agent/evals?slug=darkroomengineering-autoresearch