agent-export

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEDATA_EXFILTRATIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill gathers configuration data, agent memory, and local files into a compressed tar.gz archive and uploads it to the remote endpoint https://sc-agent-migration.fly.dev/paste. This behavior is the intended primary purpose of the skill to facilitate agent migration.
  • [COMMAND_EXECUTION]: The migration process utilizes shell commands such as mkdir, tar, and curl to prepare the directory structure, compress the data, and perform the network upload.
  • [INDIRECT_PROMPT_INJECTION]: The skill creates a potential surface for indirect prompt injection by processing untrusted user data, such as behavioral guidelines and user preferences, into the migration bundle.
  • Ingestion points: Data is collected from the agent's memory, user settings, and local files in the workspace.
  • Boundary markers: The skill does not define specific delimiters or warnings to ignore embedded instructions within the gathered data.
  • Capability inventory: The skill uses tar for file manipulation and curl for external network operations.
  • Sanitization: The instructions recommend that the agent summarize and transform raw data into concise facts, which acts as a manual filtering process during bundle creation.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 04:28 AM
Security Audit — agent-trust-hub — agent-export