twinmind-migration-deep-dive
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes transcript data from local files and external APIs without explicit sanitization or boundary markers. This untrusted data enters the agent context through file reading and API fetches. The skill possesses capabilities like shell command execution and file modification, creating a surface where malicious instructions embedded in transcripts could potentially influence agent actions.\n
- Ingestion points:
references/implementation.md(viafs.readFileSyncandfetchfrom external APIs)\n - Boundary markers: Absent\n
- Capability inventory:
SKILL.md(Bash, curl, Write, Edit)\n - Sanitization: Absent\n- [COMMAND_EXECUTION]: The skill utilizes shell commands via
curlto interact with the TwinMind API and verify enterprise configurations.\n- [EXTERNAL_DOWNLOADS]: The skill fetches transcript data from official API endpoints of Fireflies.ai and Rev.ai to facilitate the migration process.
Audit Metadata