The Agent Skills Directory

[COMMAND_EXECUTION]: The test suite (tests/test_fireflies.py) utilizes subprocess.run to execute the skill's CLI script for automated validation. This is an expected pattern for testing CLI-based tools within a development environment.
[DATA_EXPOSURE]: The skill interacts with the official Fireflies.ai GraphQL API. Authentication is handled via a FIREFLIES_API_KEY environment variable or command-line flag. The documentation provides clear guidance on secure secret management, advising against hardcoding or exposing credentials.
[EXTERNAL_DOWNLOADS]: The skill supports a remote audio upload feature (scripts/fireflies audio upload) which transmits a media URL to the Fireflies.ai service for processing. The network operation is performed by the remote service rather than the local agent environment.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes meeting transcripts and summaries which may contain untrusted data.
Ingestion points: Meeting transcripts, summaries, and AskFred responses are retrieved via scripts/fireflies and ingested into the agent context.
Boundary markers: The CLI returns data in structured JSON format, providing clear boundaries for the agent.
Capability inventory: The skill allows network requests to the official Fireflies API and mutation capabilities (renaming, deleting, or querying meeting data) which require explicit confirmation.
Sanitization: The skill outputs raw JSON from the API; it relies on standard agent platform guardrails for safe interpretation of external content.

fireflies