docs-check
Pass
Audited by Gen Agent Trust Hub on Mar 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local shell script at
scripts/fetch-claude-docs.shto automate the retrieval of documentation. The source code for this script was not included in the analyzed files. - [EXTERNAL_DOWNLOADS]: The skill uses
curlto download documentation fromcode.claude.com. These are official resources from Anthropic. - [PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it fetches external markdown files and provides them as context to subagents without sufficient sanitization. Ingestion points: External markdown files fetched from
code.claude.comand stored in.claude/docs-check/docs-cache/. Boundary markers: The provided files do not define delimiters or specific instructions to ignore embedded commands within the fetched documents. Capability inventory: The orchestrator and its subagents (Sonnet/Haiku) have capabilities to scan directories, write reports, and propose pull requests. Sanitization: No evidence of content filtering or validation was found for the markdown data before it is interpolated into agent prompts.
Audit Metadata