research Eval
=============

Status: review
Score: 78/100
Risk: medium
Decision: manual_review
Policy: review
Reason: Require human approval before installing into a real workspace.

Install:
npx skills add Weizhena/Deep-Research-skills --skill research

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: Good trust signals with a few areas worth checking before rollout.
- WARN Audit score: Needs review
- WARN Agent safety gate: Usable candidate, but the agent should surface permission and audit notes before installation.
- PASS License clarity: MIT
- PASS Permission surface: no high-risk permission surface in public metadata

Warnings:
- Trust score: Good trust signals with a few areas worth checking before rollout.
- Audit score: Needs review
- Agent safety gate: Usable candidate, but the agent should surface permission and audit notes before installation.
- README/SKILL.md completeness: Public metadata needs stronger README/SKILL.md context
- The SKILL.md does not explicitly instruct the web-search agent to treat web content as untrusted data, which could allow prompt-injection text from search results to influence the generated outline or subsequent tasks.
- The skill lacks a dedicated 'Limitations' section, so it is unclear how it should behave for very broad topics, ambiguous topics, or topics with fragmented or unreliable sources.
- Step 4 mentions merging step1_output, step2_output, and user-defined fields, but the exact merge semantics and schema for step2_output are not fully specified, which may lead to inconsistent outline.yaml and fields.yaml generation.
- 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
- The SKILL.md does not explicitly instruct the web-search agent to treat web content as untrusted data, which could allow prompt-injection text from search results to influence the generated outline or subsequent tasks.
- No OpenAgentSkill engagement data yet
- The skill lacks a dedicated 'Limitations' section, so it is unclear how it should behave for very broad topics, ambiguous topics, or topics with fragmented or unreliable sources.
- Step 4 mentions merging step1_output, step2_output, and user-defined fields, but the exact merge semantics and schema for step2_output are not fully specified, which may lead to inconsistent outline.yaml and fields.yaml generation.
- Quality score needs review
- README/SKILL.md completeness: Public metadata needs stronger README/SKILL.md context

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