The Agent Skills Directory

[DATA_EXFILTRATION]: The skill accesses the local ~/.claude/history.jsonl file, which contains private developer logs, project context, and pasted source code.
Evidence: Step 1 instructions require the agent to read ~/.claude/history.jsonl to filter developer interactions from the past 48 hours.
Evidence: Step 7 instructions utilize the Rube MCP tool to send a report containing summarized history data to an external Slack channel via RUBE_MULTI_EXECUTE_TOOL.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted historical chat data from local logs, creating a potential vector for indirect prompt injection if those logs contain malicious instructions.
Ingestion points: ~/.claude/history.jsonl accessed in Step 1.
Boundary markers: Absent. No instructions are given to the agent to use delimiters or specifically ignore instructions that might be embedded within the project chat logs.
Capability inventory: Includes network search capabilities via RUBE_SEARCH_TOOLS and external messaging via RUBE_MULTI_EXECUTE_TOOL.
Sanitization: Absent. The skill does not describe any sanitization, filtering, or validation of the history content before processing it for report generation.

developer-growth-analysis