gift-push-strategy

Pass

Audited by Gen Agent Trust Hub on Aug 24, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses ae-cli commands to interact with the Thinking Engine platform. This includes querying event metadata (ae-cli analysis-meta event list), configuration lists (ae-cli engage-scene strategy list), and submitting operations for approval (ae-cli engage-task task submit-approval). These actions are core to the skill's functionality and are protected by a mandatory user confirmation step and a batch limit of 10 tasks.
  • [DATA_EXPOSURE]: Persists project-specific context in a local .gift-push-context.md file. This is used for session memory and to maintain user preferences, such as custom tiering rules and lifecycle groupings.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources, specifically Feishu documents (via lark-wiki) and TE system queries.
  • Ingestion points: External Feishu links and TE event/property lists.
  • Boundary markers: The skill uses predefined markdown templates (e.g., references/slg.md, references/card.md) to structure the ingested data.
  • Capability inventory: File system write access (context file) and task submission via CLI tools.
  • Sanitization: Relies on specific domain constraints (game genre templates) to parse and validate incoming strategy data.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 24, 2026, 10:17 AM
Security Audit — agent-trust-hub — gift-push-strategy