The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes ripgrep (rg) and zg (a vector search tool) CLI commands to perform searches within the user's ~/ops directory. These operations are local and align with the skill's stated purpose of searching a diagnosis vault.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external markdown files which could influence the agent's behavior during a diagnosis if the files contain malicious instructions.
Ingestion points: Markdown files (*.md) and an index file (index.md) located in the ~/ops directory.
Boundary markers: No explicit delimiters or instructions are used to ignore potentially malicious embedded content within the searched files.
Capability inventory: The skill executes shell commands (rg, zg) and reads file contents into the agent's context.
Sanitization: No validation or sanitization is performed on the content retrieved from the vault files before it is processed by the agent.
Context: This attack surface is a direct consequence of the skill's intended functionality to leverage prior knowledge for bug diagnosis, and the data source is presumed to be the user's own knowledge base.

diag-search