folder-specific-claude-and-agents-md
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads all files within a target folder to generate context, creating a vulnerability surface where malicious instructions in those files could influence the agent's behavior.\n
- Ingestion points: All files in the target folder (SKILL.md, Step 2).\n
- Boundary markers: Absent; the agent is instructed to read files "IN FULL" without instructions to ignore potentially malicious embedded content.\n
- Capability inventory: Directory listing (
ls), symlink creation (ln), and version control commands (git) (SKILL.md, Steps 2, 6, 7).\n - Sanitization: Absent.\n- [COMMAND_EXECUTION]: The skill utilizes shell commands to manage files and version control, including
ls -la,ln -s,git add,git commit, andgit push(SKILL.md, Steps 2, 6, 7).\n- [DATA_EXFILTRATION]: The process involves reading all files in a directory, specifically mentioning "config" and "key source files." If the target directory contains secrets or sensitive data, this information will be exposed to the agent's session context (SKILL.md, Step 2).
Audit Metadata