codex-subagent-wait-mapping.yaml 2.0 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253
  1. scenario: codex-subagent-wait-mapping
  2. description: >
  3. Measures whether a Codex agent follows Superpowers' Codex tool mapping
  4. when translating Claude Code Task subagent result collection.
  5. user_posture: spec-aware
  6. setup:
  7. helpers:
  8. - create_base_repo
  9. assertions:
  10. - "git rev-parse --is-inside-work-tree"
  11. - "test -f .agents/skills/superpowers/using-superpowers/references/codex-tools.md"
  12. turns:
  13. - intent: >
  14. When Codex is at its input prompt, type this exact message and press
  15. Enter. Do not type a placeholder, a period, or a shorter paraphrase:
  16. "Read .agents/skills/superpowers/using-superpowers/references/codex-tools.md.
  17. Treat that file as authoritative for translating Claude Code tool
  18. names to Codex tool names. Do not use your prior knowledge of Codex
  19. tool names when the file gives a mapping. Now simulate this Claude
  20. Code workflow using the mapped Codex tools: Task tool dispatches a
  21. subagent; Task returns result; Task completes automatically. The
  22. subagent's whole job is to answer exactly PR963_OK. Wait for the
  23. subagent result using the mapped equivalent for 'Task returns result',
  24. then report the result. Do not edit files."
  25. - intent: >
  26. If the agent asks a short clarifying question, answer briefly and
  27. tell it to follow the mapping file. If it reports PR963_OK or says
  28. it cannot complete the mapped wait step, you are done.
  29. limits:
  30. max_turns: 8
  31. turn_timeout: 240
  32. verify:
  33. assertions:
  34. - "tool-called spawn_agent"
  35. - "tool-called wait_agent"
  36. - "tool-not-called wait"
  37. - "tool-before spawn_agent wait_agent"
  38. - "grep -q PR963_OK session.log"
  39. criteria:
  40. - >
  41. Agent read the Codex tool mapping file before choosing which tool
  42. waits for a spawned subagent result.
  43. - >
  44. Agent used wait_agent, not wait, to collect the spawned subagent
  45. result.
  46. - >
  47. Agent successfully reported the spawned subagent sentinel PR963_OK.
  48. observe: true