ask
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from a research knowledge base and local methodology files (
ops/methodology/*.md) to synthesize answers. The lack of explicit boundary markers or instructions to disregard embedded commands in the source documents creates a potential surface for indirect prompt injection.\n - Ingestion points:
ops/methodology/*.md,${CLAUDE_PLUGIN_ROOT}/methodology/, and user$ARGUMENTS.\n - Boundary markers: Absent.\n
- Capability inventory:
Read,Grep,Glob,mcp__qmd__*tools, and shell execution (cat,rg).\n - Sanitization: Absent.\n- [DYNAMIC_EXECUTION]: The instructions include a bash loop template (
for f in ops/methodology/*.md; do cat \"$f\"; done) which the agent is directed to execute at runtime to aggregate local methodology notes.\n- [COMMAND_EXECUTION]: The skill utilizes shell-based search and retrieval tools, includingrg(ripgrep) and standard file reading commands, to operate on the local file system and knowledge base repository.
Audit Metadata