The Agent Skills Directory

[COMMAND_EXECUTION]: The Python script in SKILL.md executes git commands (remote get-url and rev-list) via subprocess.run to retrieve project metadata and commit counts for the report. These operations are restricted to project directories found in the history file.\n- [DATA_EXFILTRATION]: The skill accesses the local file ~/.claude/history.jsonl (SKILL.md) to gather usage statistics. While this file contains sensitive interaction history, the data is processed locally to generate an HTML report and no evidence of network transmission was found.\n- [PROMPT_INJECTION]: The skill processes untrusted data from the local history file, creating a surface for indirect prompt injection.\n
Ingestion points: Reads ~/.claude/history.jsonl in SKILL.md.\n
Boundary markers: Absent; the script parses the JSONL file directly without delimiters or instructions to ignore embedded prompts.\n
Capability inventory: Uses subprocess.run for git operations and open() for writing the local HTML report (SKILL.md).\n
Sanitization: The script focuses on metadata (counts, timestamps) and does not currently render prompt content in the report, which mitigates the risk of malicious input affecting the output.

cc-analytics