Parent: #66345 — GPT 5.4 Enhancement v3: Hermes Parity Sprint

Priority: P2 — LOW
Type: Quality Polish

Problem

OpenClaw's GPT-5 execution bias section lacks a structured verification checklist. The current prompt says "Act first, then verify if needed" but doesn't specify what to verify. Hermes provides a 5-point checklist that GPT-5 uses to self-check before finalizing responses.

The critical missing dimension is grounding — explicitly checking that factual claims are backed by tool output, not training data. This reinforces PR 1's mandatory tool-use categories.

Hermes Reference

agent/prompt_builder.py lines 238-245:

"<verification>\n"
"Before finalizing your response:\n"
"- Correctness: does the output satisfy every stated requirement?\n"
"- Grounding: are factual claims backed by tool outputs or provided context?\n"
"- Formatting: does the output match the requested format or schema?\n"
"- Safety: if the next step has side effects, confirm scope before executing.\n"
"</verification>"

Proposed Implementation

File: extensions/openai/prompt-overlay.ts

Append to OPENAI_GPT5_EXECUTION_BIAS:

### Verification
Before finalizing your response, check:
- Correctness: does the output satisfy every stated requirement?
- Grounding: are factual claims backed by tool outputs, not training data?
- Coverage: did you address every part of the request?
- Formatting: does the output match the requested format or schema?
- Safety: if the next step has side effects (file writes, commands, API calls), confirm scope before executing.

Verification Flow

  GPT-5 finishes tool calls
           │
           ▼
  ┌─────────────────────┐
  │ Verification Pass    │
  │                      │
  │ ✓ Correctness        │ ← Does output match requirements?
  │ ✓ Grounding          │ ← Claims backed by tool output?  (KEY)
  │ ✓ Coverage           │ ← All parts addressed?
  │ ✓ Formatting         │ ← Right format/schema?
  │ ✓ Safety             │ ← Side effects scoped?
  │                      │
  └──────────┬───────────┘
             │
             ▼
  Finalize response

Impact

Modest improvement in answer quality, particularly for multi-step tasks. The grounding check specifically reinforces mandatory tool use — even if GPT-5 skips a tool call, the verification step prompts it to notice the gap.