The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses Bash commands to locate and parse session log files in the user's home directory. This includes using find to identify recent files and grep combined with jq to extract specific numeric fields related to API costs and token usage.
[DATA_EXPOSURE]: Accesses log files within ~/.claude and ~/.openclaw. While session logs may contain project-specific information, the skill's operations are limited to extracting cost-related metadata for reporting purposes, and no network exfiltration or credential access was observed.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external session logs which could potentially contain untrusted content.
Ingestion points: Local JSONL files found in ~/.claude/projects/ and ~/.openclaw/agents/sessions/.
Boundary markers: None identified. The skill assumes the logs follow the expected JSONL structure.
Capability inventory: Includes Bash, Read, and Write permissions, which are used to generate the final markdown report.
Sanitization: Relies on jq for structured data extraction, which filters the content to specific cost keys, but does not perform textual sanitization of the log contents.

cache-monitor