sociuu-finalize
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from several external and potentially untrusted sources which could be used to deliver malicious instructions to the agent.
- Ingestion points: The agent is directed to read and analyze content from Merge Request descriptions, ClickUp task ledgers, Figma references, and Slack/Email communications.
- Boundary markers: Absent; there are no instructions for the agent to use delimiters or ignore embedded instructions when processing external data.
- Capability inventory: The agent has the authority to delete local worktrees, prune Docker resources via 'Dock', and update external task records.
- Sanitization: Absent; the skill relies on the agent's internal logic to 'verify' proof without explicit escaping or content validation.
- [COMMAND_EXECUTION]: The skill instructs the agent to manage and delete system resources on the local host.
- Evidence: The agent is authorized to remove 'task-owned Dock environments' and 'inactive, preserved task worktrees' through workspace isolation tools. This capability could result in unauthorized data loss if the agent is manipulated by injected instructions from an untrusted external task or Merge Request.
Audit Metadata