The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes instructions to use dbus-send to query the system session bus for available application services. This is used to discover automation interfaces for desktop applications during the test infrastructure discovery phase.
[COMMAND_EXECUTION]: The skill's YAML frontmatter contains a hooks configuration that automatically executes a shell command (uv run python3) whenever Grep, Glob, or Agent tools are used. This command executes a local Python script (phase-gate-guard.py) to enforce workflow state validation.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it is designed to ingest and process untrusted data from an external codebase.
Ingestion points: Codebase content is read into the agent context via Read, grep, ast-grep, and ripgrep-all tools.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore embedded instructions' warnings when processing retrieved code snippets.
Capability inventory: The skill possesses significant capabilities, including the ability to spawn background sub-agents, execute system commands (dbus-send), and run Python scripts (uv run).
Sanitization: There is no evidence of validation or sanitization of the content retrieved from the codebase before it is processed by the agent.
[PROMPT_INJECTION]: The skill contains 'No Pause' instructions that direct the agent to immediately transition to the next workflow phase (dev-clarify) without summarizing findings, asking for user confirmation, or waiting for user input. This automation pattern reduces human oversight and could allow malicious instructions encountered during exploration to influence subsequent phases without detection.

dev-explore