The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external log files and episodic memory documents which can contain untrusted content from various communication channels.
Ingestion points: The skill reads .beta/logs/app.log and CLAUDE.md (episodic memory) to diagnose issues.
Capability inventory: The skill has access to the run_code and read_file tools to execute diagnostics.
Boundary markers: No specific delimiters or instructions to ignore embedded commands are present in the logic that processes log data.
Sanitization: There is no evidence of filtering or sanitization being applied to the data retrieved from logs or memory before it is processed by the agent.
[COMMAND_EXECUTION]: The skill executes shell commands and database queries to verify runtime health.
Evidence: Instructions to run betaclaw status, PRAGMA integrity_check, and PRAGMA journal_mode using the run_code tool.
[DATA_EXPOSURE]: The skill accesses sensitive configuration and security assets internal to the betaclaw environment.
Evidence: The instructions mandate reading .beta/config.toon and verifying that the vault is accessible and decryptable.
Context: These actions target the vendor's own application structure for the purpose of debugging.

debug