The Agent Skills Directory

[COMMAND_EXECUTION]: Provides bash command snippets utilizing well-known tools such as jq, rg, awk, and grep to process session data.
[DATA_EXFILTRATION]: Accesses the agent's internal conversation history and session logs located in the ~/.clawdbot/agents/ directory.
[PROMPT_INJECTION]: Ingests historical conversation logs which may contain untrusted content from previous user interactions, creating an indirect prompt injection surface.
Ingestion points: Reads message transcripts from .jsonl files stored in the agent's local session directory.
Boundary markers: The skill does not define explicit delimiters or instructions to ignore embedded commands when presenting historical text to the agent.
Capability inventory: Executes shell commands for searching, filtering, and summarizing log data.
Sanitization: Uses jq filtering to specifically target and extract text fields from the structured JSON messages.

session-logs