The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes local shell commands and helper scripts to collect session metrics and history.
Evidence: Executes git log, git diff, and local scripts (list-sessions.sh, extract-data.sh) to retrieve metadata and commit statistics.
[DATA_EXFILTRATION]: The skill accesses local session logs which contain project conversation history.
Evidence: Reads session data from ~/.claude/projects/ to perform its primary analysis. Generated reports are stored locally within the project's skill directory.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it processes data from previous session logs.
Ingestion points: Reads conversation logs from the local filesystem via the extract-data.sh script.
Boundary markers: No explicit delimiters are specified for the log content processing.
Capability inventory: The skill can modify project configuration files (like CLAUDE.md and SKILL.md) using the Edit tool.
Sanitization: All configuration changes are gated by a mandatory user validation step (Step 7), requiring manual review and approval before execution.

retrospecting