jbvc/search-first
Research-before-coding workflow. Search for existing tools, libraries, and patterns before writing custom code. Invokes the researcher agent.
48
60%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
View guide
Low
Low-risk findings worth noting
Low
Low-risk findings.
1 low severity finding. Worth noting, but not necessarily harmful.
W011: Third-party content exposure detected (indirect prompt injection risk).
What this means
The skill exposes the agent to untrusted, user-generated content from public third-party sources, creating a risk of indirect prompt injection. This includes browsing arbitrary URLs, reading social media posts or forum comments, and analyzing content from unknown websites.
Why it was flagged
SKILL.md's "Full Mode (agent)" explicitly instructs the researcher agent to search and ingest results from public sources (npm, PyPI, GitHub and the web/MCP servers), which are untrusted, user-generated third‑party content that the agent must read and that can materially influence tool selection and follow-up actions.
Where we found it
npmjs.com
domain · 8 sites
The Full Mode (agent) prompt at line 78 explicitly instructs the researcher agent to search npm/PyPI for packages, and the workflow diagram at line 28 lists npm as a parallel search source; the agent would fetch and read untrusted package metadata and READMEs from npmjs.com.
pypi.org
domain · 5 sites
The Full Mode (agent) prompt at line 78 explicitly instructs the researcher agent to search npm/PyPI for packages, and the workflow diagram at line 29 lists PyPI as a parallel search source; the agent would fetch and read untrusted package metadata and READMEs from pypi.org.
SKILL.md
29
│ │ PyPI │ │ Skills │ │ Web │ │
SKILL.md
63
1. Is this a common problem? → Search npm/PyPI
SKILL.md
78
Search: npm/PyPI, MCP servers, Claude Code skills, GitHub
SKILL.md
121
- Cycle 1: Broad search (npm, PyPI, MCP)
SKILL.md
139
Search: npm "http client retry", PyPI "httpx retry"
github.com
domain · 3 sites
The Full Mode (agent) prompt at line 78 explicitly instructs the researcher agent to search GitHub, and the workflow diagram at line 28 lists GitHub as a parallel search source; the agent would fetch and read untrusted repository content, issues, and READMEs from github.com.
SKILL.md
28
│ │ npm / │ │ MCP / │ │ GitHub / │ │
SKILL.md
66
4. Is there a GitHub template? → Search GitHub
SKILL.md
78
Search: npm/PyPI, MCP servers, Claude Code skills, GitHub
MCP servers
content-type · 8 sites
The Full Mode (agent) prompt at line 78 explicitly instructs the researcher agent to search MCP servers, and the workflow diagram at line 28 lists MCP as a parallel search source; the agent would fetch and read untrusted content from arbitrary third-party MCP servers.
Claude Code skills
content-type · 1 site
The Full Mode (agent) prompt at line 78 explicitly instructs the researcher agent to search Claude Code skills, which are user-generated third-party skill definitions that the agent would read and could influence tool selection.
SKILL.md
78
Search: npm/PyPI, MCP servers, Claude Code skills, GitHub
Web search
content-type · 1 site
The workflow diagram at line 29 lists "Web" as a parallel search source in the third column alongside GitHub and Skills, indicating the researcher agent is instructed to search the open web for untrusted third-party content.
SKILL.md
29
│ │ PyPI │ │ Skills │ │ Web │ │
- Audited
- Security analysis