RUN CHECKLIST

GPTClaudeGemini··▲ 1,280 copies·updated 2026-07-14

run-checklist.prompt

# TL17 Prompt Sweep Checklist (Standardized)

Use this flow for any timeline topic. The active prompt set is:
- fixed/no-K: `pA`, `pB`, `pC`, `pD`, `pE`, `pF`
- helping/with-K: `pK`

## 1) Choose run settings
- `TOPIC`: folder under `samples/` (example: `msft_linkedin`)
- `QUERY_TITLE`: timeline title used as query
- `MODEL`: model name
- `PROMPT_ID`:
  - no-K condition: `pA`, `pB`, `pC`, `pD`, `pE`, `pF`
  - with-K condition: `pK`
- `TAG`: short run id, e.g. `gpt-5.2_pA_t0`

Only for with-K (`pK`):
- `K`: target entry count (typically number of reference date blocks)

## 2) Generation protocol
- Query-only generation (no external retrieval/context text).
- Use one model at a time to isolate provider failures.
- Use request timeout in generation.

### 2a) No-K condition (fixed prompts)

when to use it

Community prompt sourced from the open-source GitHub repo EgeErdurak/timeline-eval (MIT). A "RUN CHECKLIST" style prompt — adapt the placeholders and specifics to your task. Imported as-is and not independently retested here, so check the output before relying on it.