codex-gateway
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use the 'tempo' CLI tool for wallet management (e.g., 'tempo wallet login', 'tempo wallet -t whoami') and for executing paid GraphQL requests ('tempo request'). These are standard operations for interacting with the codex-data platform.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes data returned from the Codex GraphQL API, creating a potential surface for indirect prompt injection if the API response contains malicious instructions.\n
- Ingestion points: GraphQL query responses from 'https://graph.codex.io/graphql' as specified in 'SKILL.md'.\n
- Boundary markers: The skill does not define specific boundary markers or instructions to treat the API response data as untrusted content.\n
- Capability inventory: The agent can execute shell commands via the 'tempo' CLI for network and wallet operations (documented in 'rules/wallets.md').\n
- Sanitization: No explicit sanitization or validation of the GraphQL response data is mentioned in the skill files.
Audit Metadata