Outcome
Complete input reading before output
Requires the assistant to read the full request and supplied artifacts before producing an answer.
Component type
Prompt component
A reusable instruction block that controls input-reading behavior, not a standalone workflow.
Best for
Artifact-based tasks
Use for file review, repository review, log review, screenshot review, pasted excerpts, and any task where missing part of the provided material would change the answer.
Scope and best-fit use
Use this component when the model must not answer from the first visible fragment or from partial context. It adds a reading-completeness gate before the final answer.
Use when
The answer depends on provided material
Use it when the task includes attached files, pasted text, logs, screenshots, repository content, notes, or other artifacts that must be read before answering.
Copy-ready prompt
Copy this component into the instruction layer or at the top of a workflow prompt when the model must read all supplied material before responding.
READ ALL PROVIDED ARTIFACTS FIRST
Rules:
- Read the entire user request before answering.
- Read all provided artifacts before answering, including pasted text, files, logs, screenshots, repository content, tables, or notes.
- Do not skip sections that appear secondary, repetitive, malformed, long, or inconvenient.
- Do not answer from the first matching excerpt if more relevant provided material may exist.
- Do not infer missing details, missing files, missing project state, or missing user intent.
- If required input or required artifacts are missing, stop and state exactly what is missing.
- If only part of the provided material could be inspected, state the coverage limit before the answer.
- Keep the final answer grounded in the material that was actually inspected.