developer-growth-analysis
Fail
Audited by Gen Agent Trust Hub on Aug 6, 2026
Risk Level: HIGHDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill reads the Claude Code history file located at
~/.claude/history.jsonl. This file contains sensitive information, including all past user messages, source code snippets, and potential secrets or project-specific data. This data is then summarized and transmitted to Slack, an external service, which constitutes the movement of sensitive local data to a remote platform. - [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by ingesting untrusted data from the
history.jsonlfile (specifically thedisplayandpastedContentsfields) without any boundary markers or sanitization logic. A malicious instruction previously present in the chat history could be interpreted as a command during the analysis or report generation phases. - Ingestion points:
~/.claude/history.jsonl(File read operation). - Boundary markers: None identified; the skill directly interpolates history contents into its analysis prompt.
- Capability inventory: Sends messages to Slack via
RUBE_MULTI_EXECUTE_TOOL, performs external searches viaRUBE_SEARCH_TOOLS. - Sanitization: None identified; no filtering or escaping is performed on the extracted chat content before processing.
Recommendations
- AI detected serious security threats
Audit Metadata