qmd
Pass
Audited by Gen Agent Trust Hub on Mar 5, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill installs the
qmdutility from a public GitHub repository (https://github.com/tobi/qmd) as part of its setup process. - [COMMAND_EXECUTION]: The skill relies on executing local CLI commands to manage collections and perform searches, such as
qmd collection add,qmd update, andqmd query. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it indexes and retrieves content from local files. If an indexed file contains malicious instructions, these could be ingested by the agent during a search and potentially influence its subsequent actions.
- Ingestion points: Local files indexed via the
qmd collection addcommand as specified inSKILL.md. - Boundary markers: No boundary markers or "ignore instructions" delimiters are defined for the retrieved file content.
- Capability inventory: File system indexing and retrieval through the
qmdbinary. - Sanitization: No content sanitization or validation is implemented before data is presented to the agent.
Audit Metadata