The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill includes a Python script template embedded within a shell block to analyze local JSONL session logs. The script is self-contained, uses standard libraries to calculate record sizes, and does not perform network operations or dynamic code evaluation of the parsed data.
[COMMAND_EXECUTION]: A shell command template is provided to execute the Python diagnostic script. This is used for local file analysis and does not involve elevated privileges or suspicious command chains.
[INDIRECT_PROMPT_INJECTION]: The skill defines a process for ingesting external data from local session log files.
Ingestion points: Local JSONL files selected by the user and processed by the diagnostic script in SKILL.md.
Boundary markers: None; the script processes the file content line-by-line.
Capability inventory: The script is limited to reading files and printing aggregate byte counts to standard output. It lacks network, file-write, or shell-execution capabilities.
Sanitization: The script uses json.loads() for parsing and includes exception handling for encoding or JSON format errors.
[SAFE]: External references target official developer documentation for recognized well-known services (OpenAI Codex and Anthropic Claude Code).

context-budget