The Agent Skills Directory

[SAFE]: The skill follows security best practices for administrative integrations, including mandatory secret management for the FIREFLIES_API_KEY, prohibition of logging sensitive credentials, and explicit requirements for human approval before any mutations are executed. It includes specific error handling to prevent privilege escalation and ensure the preservation of administrative accounts.\n- [DATA_EXFILTRATION]: Instructions mandate that API keys must be injected from a secret manager, used only in authorization headers, and never exposed in logs, artifacts, or URLs. Redacted receipts are required to ensure no meeting-derived content enters system logs.\n- [EXTERNAL_DOWNLOADS]: The skill references official documentation and GraphQL schema endpoints from docs.fireflies.ai, which is the well-known service provider for this integration. These references are used for validating operation contracts and ensuring compliance with platform limits.\n- [INDIRECT_PROMPT_INJECTION]: The skill interacts with external API data from Fireflies.ai, presenting a potential surface for indirect injection that is mitigated by operational constraints.\n
Ingestion points: Fireflies.ai GraphQL API responses (transcripts, user data, channels) as described in SKILL.md and references/official-docs.md.\n
Boundary markers: Instructions require the agent to record redacted receipts and explicitly forbid including meeting-derived content in general output.\n
Capability inventory: Utilizes standard repository management tools including Read, Glob, Grep, Write, and Edit.\n
Sanitization: The skill enforces validation checks before reporting success and mandates that no secret or meeting-derived content enters logs or artifacts.

fireflies-enterprise-rbac