Draft content from evidence
Read target prompts, observed queries, and cited pages, then draft the amount of content the task requires.
When to use this workflow
“Write an FAQ in the page’s language to address the gaps we found.”
Before fetching
Prepare the page text, publication language, audience, approved product facts, and target gaps. Decide whether the task needs a full draft, FAQ, or focused patch.
Complete the CLI quickstart and confirm the workspace with orc workspace current --json. Replace example IDs, URLs, dates, and settings. Apply only approved mutations.
1. Fetch the questions the draft must answer
Fetch the prompt and observed search queries. Use competitor pages to understand coverage, not as evidence of facts about your own product.
orc prompts get YOUR_PROMPT_ID --json
orc fanout-queries list --type search --prompt-id YOUR_PROMPT_ID --json
orc answers list --prompt-id YOUR_PROMPT_ID --json
orc answers get YOUR_ANSWER_ID --json2. Write the required scope
Have the agent read the evidence and current page, then draft in the publication language. Tie numbers and specifications to approved sources. Preserve existing text when a focused patch is enough.
3. Prepare the draft for review
Deliver the draft as a local file with a note linking edits to the gaps they address. Publishing or writing to the site is a separate action.
Deliverable
Deliver the draft, change scope, prompt IDs, factual sources, and open review questions.
On fetch failure, retain the error and request scope. Do not count empty data, incomplete pagination, or missing permissions as a completed investigation.
Next steps
If a failure remains, report it and verify the fix, including this workflow and the failed step.