wiki-knowledge-agent Eval
=========================

Status: failed
Score: 59/100
Risk: high
Decision: do_not_auto_install
Policy: block
Reason: Trust score: Potentially useful, but at least one trust signal needs human inspection.

Install:
npx skills add atukunare/wiki-knowledge-agent --skill wiki-knowledge-agent

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
- FAIL Trust score: Potentially useful, but at least one trust signal needs human inspection.
- WARN Audit score: Needs review
- FAIL Agent safety gate: This skill should not be selected by an agent without explicit human security review.
- PASS License clarity: MIT
- FAIL Permission surface: secrets or environment access, shell or command execution

Warnings:
- Audit score: Needs review
- High-risk permission hints: Shell or command execution, Secrets or environment access
- Dependency or permission surface needs review
- Permission surface may require sandboxing
- SKILL.md does not explicitly address prompt-injection risks from pasted or fetched content, even though the skill allows the LLM to classify, summarize, and optionally update config.
- The skill instructs fetching user-supplied URLs but does not define safe URL boundaries, such as blocking private/internal IP ranges or stripping credentials from URLs.
- The excerpted skill directory does not clearly include referenced scripts (scripts/ingest.py, scripts/onboarding.py) and reference files; confirm they are present in the repository or the SKILL.md pointers will be broken.
- SKILL.md lacks a dedicated limitations/safety section describing what the skill will not do, such as refusing instructions embedded in saved content or requiring confirmation before sending notifications.
- Low GitHub adoption signal
- Quality score needs review
- Permission surface needs review: secrets or environment access, shell or command execution
- GitHub adoption: 18 GitHub stars

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 does not explicitly address prompt-injection risks from pasted or fetched content, even though the skill allows the LLM to classify, summarize, and optionally update config.
- High-risk permission hints: Shell or command execution, Secrets or environment access
- Dependency or permission surface needs review
- Permission surface may require sandboxing
- The skill instructs fetching user-supplied URLs but does not define safe URL boundaries, such as blocking private/internal IP ranges or stripping credentials from URLs.

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