project-indexer
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill scans the project's file structure, reads file contents to extract symbols, and generates one-line descriptions for every file. This constitutes an ingestion point for untrusted data if the codebase contains malicious instructions intended to hijack the agent's behavior during the indexing process.
- Ingestion points: The indexing workflow in
SKILL.md(Step 2: Scan and Analyze Project) involves traversing the file structure and reading file contents to extract symbols and generate descriptions. - Boundary markers: The instructions do not specify the use of delimiters or clear "ignore embedded instructions" warnings when processing the contents of project files.
- Capability inventory: The skill has the capability to write and modify files in the project root (
CLAUDE.md) and create new directories/files (.claude-index/index.md,.claude-index/config.md). - Sanitization: There is no evidence of sanitization or filtering of the extracted code symbols or file content before they are used to generate the index or updated in
CLAUDE.md.
Audit Metadata