cookbook-audit
Audit an Anthropic Cookbook notebook based on a rubric. Use whenever a notebook review or audit is requested.
74
61%
Does it follow best practices?
Impact
100%
1.96xAverage score across 3 eval scenarios
Low
Low-risk findings worth noting
Fix and improve this skill with Tessl
tessl review fix ./.claude/skills/cookbook-audit/SKILL.mdThe canonical home for this skill is cookbook-audit in anthropics/claude-cookbooks
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
The required workflow runs `python3 validate_notebook.py <path>` which reads the user-supplied notebook JSON and converts it to markdown (`nbconvert --to markdown`), then the generated markdown is used for “Manual review” (i.e., fed into the agent’s LLM context as review text), so outsider-authored notebook content can reach the LLM.
- Repository
- anthropics/anthropic-cookbook
- Commit
d7265d6
- Audited
- Security analysis
Is this your skill?
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.