The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses several system utilities to gather diagnostics. Specifically:
It invokes journalctl --user and systemctl --user to monitor the clawdbot-gateway.service. These commands operate within the user's permission scope and do not trigger privilege escalation.
It utilizes jq to parse configuration and session files located in ~/.clawdbot/.
It executes local shell scripts (scripts/response-times.sh, scripts/session-stats.sh) to process log data and calculate performance metrics.
[DATA_EXPOSURE]: The skill accesses application-specific files that may contain sensitive information:
~/.clawdbot/clawdbot.json: Analyzes the bot configuration.
~/.clawdbot/agents/main/sessions/*.jsonl: Reads conversation logs and token usage data.
These accesses are consistent with the skill's stated purpose of debugging and diagnostics for the Clawdbot application.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from log files that contain previous user messages and bot responses.
Ingestion points: Reads from ~/.clawdbot/agents/main/sessions/*.jsonl and /tmp/clawdbot/*.log using tail and jq.
Boundary markers: None identified; logs are processed as raw text or JSON.
Capability inventory: The agent can execute service management commands (systemctl) and log retrieval (journalctl).
Sanitization: There is no explicit sanitization of the log content before it is presented to the agent for analysis, which presents a surface for indirect prompt injection if the logs contain malicious instructions.

clawdbot-logs