The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a Python script using the shell to process session history.
Evidence: python3 ~/.claude/skills/history-session/scripts/history_session.py "$PWD" in SKILL.md.
[DATA_EXPOSURE_&_EXFILTRATION]: The skill accesses files within ~/.claude/projects/, which contain historical session data, including summaries and git branch names. This is the intended purpose of the skill to help users resume work.
Evidence: scripts/history_session.py reads sessions-index.json and .jsonl files from the user's home directory.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from session logs (.jsonl files) and the sessions-index.json file, displaying it back to the agent.
Ingestion points: read_jsonl and read_index functions in scripts/history_session.py.
Boundary markers: None identified in the script output; the agent is instructed to display output "exactly as-is".
Capability inventory: Shell command execution to list sessions.
Sanitization: Summaries are truncated to 45 characters, which provides minimal protection against long injection strings, but no explicit sanitization of content is performed.

history-session