developer-growth-analysis
Warn
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: MEDIUMDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [DATA_EXFILTRATION]: The skill is designed to read the contents of
~/.claude/history.jsonl, a local file containing the user's entire conversation history, project names, and pasted code. This sensitive information is then summarized and transmitted to Slack via an external tool, establishing a network data flow from local private logs to a remote service. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted historical data from past chat sessions to identify patterns. Because it lacks sanitization or boundary markers, malicious instructions or payloads contained within the history (e.g., in
pastedContents) could be re-activated or influence the agent's behavior during the analysis and reporting phase. - Ingestion points: Reads from
~/.claude/history.jsonlas defined in the 'Access Chat History' section ofSKILL.md. - Boundary markers: None; the instructions do not implement delimiters or 'ignore' commands for the content extracted from the history file.
- Capability inventory: The skill has the ability to send formatted messages to Slack using
RUBE_MULTI_EXECUTE_TOOLand perform web searches usingRUBE_SEARCH_TOOLS. - Sanitization: None; the skill interpolates historical data directly into its analysis logic without filtering for executable instructions.
- [UNVERIFIABLE_DEPENDENCIES]: The skill relies on 'Rube MCP' for searching HackerNews, managing authentication, and sending reports to Slack. These tools (
RUBE_SEARCH_TOOLS,RUBE_MANAGE_CONNECTIONS,RUBE_MULTI_EXECUTE_TOOL) are external dependencies that handle sensitive credentials and network operations but are not from a recognized trusted vendor list.
Audit Metadata