Source Intake

GPTClaudeGemini··▲ 1,230 copies·updated 2026-07-14

source-intake.prompt

# Source Intake Prompt Policy

Stage id: `source_intake`
Owner: RedCube AI
Purpose: freeze source truth, audience constraints, source gaps, and allowed evidence before communication planning.

Canonical policy:
- Read all supplied source material before producing source-readiness inputs for communication planning.
- Treat `source_materials_full_text`, source package refs, and explicit user constraints as the only source-truth basis.
- Return source truth refs and evidence-gap diagnostics. Insufficient source material is quality debt and cannot block the next stage.
- Do not write visual truth, review/export verdicts, artifact bodies, memory bodies, or OPL runtime state.

Detailed prompt locators:
- `ppt_deck`: source-readiness artifacts and source package refs only; storyline starts in `communication_strategy`.
- `xiaohongshu`: `prompts/xiaohongshu/research.md`; storyline starts in `communication_strategy`.
- `poster_onepager`: source-readiness artifacts and source package refs only; storyline starts in `communication_strategy`.

Authority boundary:
- AI-first source readiness judgment belongs to RCA stage artifacts.
- Programmatic code may validate refs and emit diagnostics; typed blockers are reserved for authority, safety, human-decision, irreversible-action, executor, or identity/currentness boundaries.

when to use it

Community prompt sourced from the open-source GitHub repo gaofeng21cn/redcube-ai (Apache-2.0). A "Source Intake" 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.