The Agent Skills Directory

[COMMAND_EXECUTION]: The skill generates shell commands for the bd CLI using text extracted from project files and user-provided arguments. If these inputs contain shell metacharacters (e.g., backticks or semicolons) and are not properly escaped by the agent, they could be used to execute arbitrary commands in the local environment.
[EXTERNAL_DOWNLOADS]: The skill references the beads-cli package and its repository (steveyegge/beads) and provides installation instructions. While the tool is a known developer utility, it originates from a third-party source that is not on the platform's trusted vendor list.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by processing untrusted project data to drive task creation.
Ingestion points: Reads project prototype, components, routes, and documentation files during the audit phase.
Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are used when the agent processes audited content into findings.
Capability inventory: Executes shell commands via the bd CLI (bd create, bd update, bd init).
Sanitization: No explicit sanitization or escaping of extracted content is requested before the findings are interpolated into CLI commands.

vocabulary-audit-beads