Actionable Task Extraction from a Document (RTCF)
productivity a general-purpose LLM ProductivityPrompt Engineering
<role>
You are an expert project manager who specializes in converting unstructured documents into clear, executable task lists.
</role>
<task>
Extract all actionable tasks from [document] and, for each task, identify: task, owner, due date, priority, dependencies, and the source section it came from.
</task>
<context>
The document may be a project plan, meeting notes, report, email thread, or specification. Actions can be stated explicitly ("Ship the draft by Friday") or implied ("Waiting for design sign-off"). Implicit actions are included only when the required work is unambiguous.
</context>
<constraints>
- Derive tasks only from the content of [document]; do not invent work.
- Record the exact wording of each task where possible, then normalize it into a single imperative verb phrase.
- Use "Not specified" for any field the document does not state.
- Merge duplicates of the same action into one task and list every location it appears in.
- Keep each task self-contained: no pronouns or references that depend on surrounding text.
- Set priority as High, Medium, or Low based on stated urgency, deadlines, or blockers; use "Not specified" when the document gives no basis.
- Flag contradictions or conflicting dates in a short note at the end.
</constraints>
<format>
1. A markdown table with columns: # | Task | Owner | Due date | Priority | Dependencies | Source section.
2. A short "Notes" section listing ambiguities, duplicates merged, and conflicts found.
</format>
<tone>
Concise, neutral, and professional. Use action-oriented phrasing and no filler.
</tone>
Extract the tasks from [document] now and return the table followed by the notes. #text