jbvc/planning-with-files
Implements Manus-style file-based planning for complex tasks. Creates task_plan.md, findings.md, and progress.md. Use when starting complex multi-step tasks, research projects, or any task requiring >5 tool calls.
13
17%
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
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 skill's SKILL.md and examples.md explicitly instruct the agent to perform web searches (e.g., "WebSearch 'morning exercise benefits'"), to read/browser results and record URLs in findings.md (templates/resources and reference.md also reference external URLs), so untrusted public web content is fetched and used to drive planning and decisions.
Where we found it
arbitrary public web content via WebSearch and WebFetch
content-type · 4 sites
W011 requires the plugin itself to fetch or read a specific external source at runtime; here the plugin only lists tool permissions in frontmatter and provides workflow planning instructions, neither of which constitutes a plugin-initiated runtime fetch of a concrete external host or URL.
SKILL.md
13
- WebFetch
SKILL.md
14
- WebSearch
SKILL.md
89
"After every 2 view/browser/search operations, IMMEDIATELY save key findings to text files."
SKILL.md
150
| Browser returned data | Write to file | Screenshots don't persist |
- Audited
- Security analysis