The Agent Skills Directory

[DATA_EXFILTRATION]: The skill is configured to read the local history file ~/.claude/history.jsonl. This file contains the complete record of user interactions, including code snippets and sensitive project data in the pastedContents field. The skill then processes this information into a report that is transmitted to an external Slack workspace via the 'Rube MCP' toolset. While this is the primary functionality of the skill, it represents a significant data exposure risk if the logs contain sensitive secrets or if the delivery target is misconfigured.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests and processes untrusted data from historical chat logs and includes it in the final report without sanitization. Ingestion points: Reads the display and pastedContents fields from ~/.claude/history.jsonl (Instruction 1). Boundary markers: Absent. There are no instructions or delimiters provided to prevent the agent from following malicious instructions that might be embedded in the historical chat data or code snippets. Capability inventory: Ability to send data to Slack and search HackerNews via Rube MCP. Sanitization: Absent. The skill does not perform any escaping or validation of the content extracted from the history logs before using it to generate the report or sending it to Slack.

developer-growth-analysis