tessl-labs/best-practice-skill-improver
Eval-driven process for improving best-practice skills — analyse eval results, research what agents get wrong, rewrite for maximum uplift, and measure improvement with scenarios.
67
84%
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
Yes — Phase 3 ("Research High-Uplift Practices") explicitly instructs the workflow to search and gather public GitHub issues, blog posts, FAQs/wikis and other external documentation and include those references so the agent reads and uses untrusted, user-generated web content to decide which practices to teach (SKILL.md "Finding candidates" and "Gathering references").
Where we found it
github.com
domain · 4 sites
The plugin instructs the agent to search and read GitHub issues (user-generated content on github.com) to identify practices, then incorporate those findings into the skill output.
SKILL.md
73
Search **GitHub issues** on the library's repo — issues with many comments or reactions are pain points the community hits repeatedly
SKILL.md
91
**GitHub issues** — where the community discusses the problem
SKILL.md
126
**Add a References section** at the bottom with links to docs, issues, and posts.
SKILL.md
401
Without references, practices look like opinions. With links to official docs and community issues, they're verifiable facts.
blog posts
content-type · 3 sites
The plugin instructs the agent to search for and read arbitrary blog posts from the web, which are untrusted third-party content that could contain prompt injection.
SKILL.md
74
Search for **"common mistakes" or "gotchas"** blog posts
SKILL.md
92
**Blog posts** — that explain the "why" with data or examples
SKILL.md
126
**Add a References section** at the bottom with links to docs, issues, and posts.
FAQ or wiki
content-type · 1 site
The plugin instructs the agent to read library FAQ/wiki pages, which are community-editable untrusted content that could contain prompt injection.
SKILL.md
75
Check the library's **FAQ or wiki** — FAQ items exist because people keep asking
official documentation
content-type · 2 sites
The plugin instructs the agent to fetch and read official documentation from external sources, which is third-party content ingested at runtime.
SKILL.md
90
**Official documentation** link — the authoritative source
SKILL.md
401
Without references, practices look like opinions. With links to official docs and community issues, they're verifiable facts.
- Audited
- Security analysis