/scout

Scout LLM Fix Prompt Generation

Purpose: Scout-specific action verbs, suppression cases, template fields for the ## LLM Fix Prompt block at the end of every Scout investigation report. Read when: You are writing the ## LLM Fix Prompt block for a Scout report, choosing an action verb, or deciding whether to suppress.

Universal authoring rules and prompt structure: _common/LLM_PROMPT_GENERATION.md. This file documents only Scout-specific verbs, suppression cases, template fields.

Scout Action Verbs

Compact reference (SKILL.md keeps only this row set; full "when/receiving" detail below): FIX, FIX-WITH-TEST, MITIGATE, INVESTIGATE-FURTHER, REFACTOR-FIX.

Each fix prompt declares one verb at the top of # Your task.

Verb Selection Heuristic

Confidence == HIGH ─┬─ scoped fix, no escalation needed ──→ FIX
                    ├─ regression test bundled with fix ──→ FIX-WITH-TEST
                    └─ structural refactor required ─────→ REFACTOR-FIX

Confidence == MEDIUM ─→ INVESTIGATE-FURTHER (or FIX with verification gate)

Confidence == LOW ───→ INVESTIGATE-FURTHER (always)

Root cause out of scope / blocked ──→ MITIGATE

Tiebreakers:

• If Scout escalates to Sentinel, do not emit a fix prompt — see suppression below.
• If the bug spans 4+ files and reflects a design issue, prefer REFACTOR-FIX over FIX.
• If reproduction failed but root cause is strongly hypothesized (MEDIUM evidence), prefer INVESTIGATE-FURTHER over FIX to force the receiving LLM to confirm the symptom first.

Scout-Specific Suppression Cases

Universal cases live in _common/LLM_PROMPT_GENERATION.md. Scout adds:

Per-Bug Fix Prompt Template (Scout Fields)

Scout adds three Scout-specific blocks on top of the universal skeleton:

• Reproduction — numbered repro steps + Expected/Actual + verbatim error/log signature
• Ruled-out hypotheses — Scout's documented dead ends with eliminating evidence
• Scout-specific anti-actions in What NOT to do (e.g., do not refactor unrelated code)

## LLM Fix Prompt

```text
# Your task
<VERB> the bug described below.

# Bug context
- Title: [brief description]
- Severity: [Critical | High | Medium | Low]
- Reproducibility: [Always | Sometimes | Rare]
- Confidence: [HIGH | MEDIUM | LOW] (Scout's diagnostic confidence)

# Root cause
[Cause description — what condition triggers the bug]

Location: `<file>:<line>` in `<function>()`

# Reproduction
1. [Step 1]
2. [Step 2]

Expected: [what should happen]
Actual: [what actually happens]

Exact error / log signature (search for this string):
```
[verbatim error message or log line]
```

# Recommended fix
Approach: [high-level fix strategy]
Files to modify: [list with expected change per file]
Constraints:
- [side effect / backward-compat note]

# Acceptance criteria
- [ ] Reproduction steps above no longer trigger the failure
- [ ] Regression test added covering the failure condition
- [ ] No new test failures in the affected module

# Ruled-out hypotheses (do not revisit)
- [hypothesis 1] — eliminated because [evidence]
- [hypothesis 2] — eliminated because [evidence]

# What NOT to do
- Do not silence the symptom (catch-and-ignore, swallow logs, suppress the error)
- Do not expand scope beyond the cited files unless evidence demands it
- Do not change unrelated code paths in the same commit
```

For INVESTIGATE-FURTHER, replace "Recommended fix" with "Verification plan". For MITIGATE, add "Root cause status". For REFACTOR-FIX, add "Structural concern".