skills/matrixy/agent-browser/slack/Gen Agent Trust Hub

slack

Pass

Audited by Gen Agent Trust Hub on Mar 29, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it ingests untrusted data from Slack conversations.\n
  • Ingestion points: Slack messages, threads, and search results (documented in SKILL.md and references/slack-tasks.md).\n
  • Boundary markers: Absent; the skill instructions do not define delimiters or specific safety warnings for the agent when processing external Slack content.\n
  • Capability inventory: The skill utilizes the agent-browser tool for UI interaction and data extraction; shell access is restricted to specific commands via frontmatter configuration.\n
  • Sanitization: Absent; no sanitization or filtering of message content is implemented before processing.\n- [SAFE]: All external interactions target official Slack domains (app.slack.com) and use standard browser automation patterns.\n- [SAFE]: No hardcoded credentials, unauthorized data exfiltration paths, or obfuscated scripts were detected in the skill components.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 29, 2026, 02:13 PM