slack-messaging
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes message content from external Slack channels and threads, which serves as a vector for potentially malicious instructions to enter the agent's context.
- Ingestion points: External data is ingested through the
slack_read_channel,slack_read_thread, and various search tools mentioned in SKILL.md. - Boundary markers: The skill defines specific 'Session Management Rules' to isolate conversations, instructing the agent to start new sessions or use subagents for unrelated top-level channel messages.
- Capability inventory: The agent can send, draft, and schedule messages to any resolved Slack channel or user.
- Sanitization: While explicit character filtering is not mentioned, the skill enforces a mandatory 'Approval Before Sending' workflow for all outgoing content, ensuring that any action taken as a result of an injection is reviewed by a human.
Audit Metadata