continual-learning

Pass

Audited by Gen Agent Trust Hub on Mar 14, 2026

Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill accesses local file system paths containing sensitive user interaction data. It reads from ~/.cursor/projects/<workspace-slug>/agent-transcripts/, which contains histories of past conversations that may include sensitive project or personal information.
  • [PROMPT_INJECTION]: Indirect Prompt Injection vulnerability surface detected. The skill processes untrusted historical chat data to generate long-term memory in AGENTS.md.
  • Ingestion points: Conversation logs located in ~/.cursor/projects/<workspace-slug>/agent-transcripts/.
  • Boundary markers: Absent. The skill does not define specific delimiters or instructions to prevent extracted text from being interpreted as instructions by the agent in future sessions.
  • Capability inventory: The skill has capabilities to read and write files locally, including the persistent memory file AGENTS.md.
  • Sanitization: While the skill provides high-level instructions to exclude secrets and sensitive data, it lacks structural sanitization or escaping to prevent the promotion of malicious instructions from transcripts into persistent memory.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 14, 2026, 12:15 AM