draft-docs
Pass
Audited by Gen Agent Trust Hub on May 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill constructs shell commands using variables derived from user-provided inputs.
- Evidence: Commands such as
mkdir -p docs/drafts,mv docs/drafts/{slug}.md {destination}/{slug}.md, andls -la docs/drafts/{slug}.mduse placeholders like{slug}and{destination}. These are populated based on the "Topic prompt" and user selection, which could lead to path traversal or command injection if the input contains malicious shell metacharacters. - [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through the analysis of untrusted codebase content.
- Ingestion points: The skill ingests content from the local codebase and
git diffoutput inSKILL.md(Step 0 and Step 3). - Boundary markers: There are no explicit delimiters or instructions provided to the agent to disregard instructions embedded within the source code being analyzed.
- Capability inventory: The skill possesses file system access (read/write/move) and shell execution capabilities (
git,mkdir,mv). - Sanitization: No sanitization or validation of the ingested code content or the generated slug is specified before being used in commands or prompts.
Audit Metadata