The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute multiple shell commands using the qmd utility to perform searches and verify status.
Evidence: qmd status, qmd query "<query>" --json, qmd vsearch "configuration" --json in SKILL.md.
[EXTERNAL_DOWNLOADS]: The skill documentation encourages the installation of an external dependency from a third-party source.
Evidence: bun install -g @tobilu/qmd referenced in both README.md and SKILL.md.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from local document indexes without explicit sanitization or boundary markers.
Ingestion points: Data enters the agent's context through the output of qmd query and qmd vsearch as described in SKILL.md.
Boundary markers: None. The instructions do not specify delimiters or warnings to treat the search results as untrusted content.
Capability inventory: The agent has the capability to execute subprocesses (qmd) and read files based on the paths returned by the search.
Sanitization: None. The agent is instructed to "Parse the JSON output and present" the results directly.

rag-search