# Draft Review Revise Loop > Purpose: a meta-skill that coordinates the three modes of writing work — generating, Parent: [Writing and Documentation](https://llms-explorer.com/tree/writing-and-documentation/) · 15 facets · 69 facts · page: https://llms-explorer.com/tree/draft-review-revise-loop/ ## Draft → Review → Revise Loop - Purpose: a meta-skill that coordinates the three modes of writing work - generating, evaluating, and improving - in discrete passes with explicit handoff points. The loop is not infinite. It has stop conditions. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#draft-review-revise-loop) - When invoked by Claude: produce a structured pass log for each iteration - findings by severity, actions taken, and which stop condition ended the loop. Do not silently revise without showing the iteration record. If the user provides a document, run the loop against it; if they describe a document, ask for the text before proceeding. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#draft-review-revise-loop) - Output shape per iteration: — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#draft-review-revise-loop) - Review findings list - each finding on its own line with severity label: [Critical], [High], [Medium], or [Low], plus a one-line description — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#draft-review-revise-loop) - Revision summary (what changed, why, keyed to finding IDs) — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#draft-review-revise-loop) - Stop-condition check (which condition applies, or "continuing to iteration N+1") — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#draft-review-revise-loop) - After completing each loop, re-read the output and confirm it addresses the user's stated goal before delivering it. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#draft-review-revise-loop) ## The Three-Pass Discipline - Writing fails in two ways: shipping too early (underdeveloped argument, missing structure) or never shipping (infinite-polish). The three-pass loop prevents both by separating incompatible cognitive modes and capping iterations before they become procrastination. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#the-three-pass-discipline) ## Pass 1 — Draft - Mode: generative. Your job is to get ideas out, not to write well. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-1-draft) - No editing while drafting. Resist fixing sentences mid-flow. Let ugly prose stand. Anne Lamott calls this the "shitty first draft" - a permission slip to write badly so you can write at all. The draft's only job is to exist. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-1-draft) - Mark gaps, don't plug them mid-draft. Use [TK] (to come) or [GAP: describe what's missing] for unknowns. Stopping to research kills momentum. Fill gaps in Pass 3. Any [TK] or [GAP:] still present at the end of Pass 3 is a Critical finding - the document is not shippable until all markers are resolved or explicitly scoped out. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-1-draft) - Time-box. One Pomodoro - 25 minutes - per major section. If the clock runs out, stop mid-sentence and move on. Unfinished prose is not a crisis; it is raw material. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-1-draft) - Work from an outline, not from instinct. Even a three-bullet skeleton prevents the most common draft failure: a conclusion that argues against the introduction. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-1-draft) - See the Skill Handoff Map at the end of this document for the full per-phase routing table. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-1-draft) ## Pass 2 — Review - Mode: evaluative. You are now a reader, not the author. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-2-review) - Read the whole thing once before commenting. No inline notes on the first pass. You need to know where the document goes before you judge any sentence in it. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-2-review) - Separate findings by type: — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-2-review) - Missing: a claim is made but the evidence is absent; a step is assumed but not stated; a section promised in the introduction does not appear. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-2-review) - Wrong: a fact is incorrect; an argument contradicts itself; a structural pattern is violated. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-2-review) - Could be better: prose is clear enough but could be sharper, shorter, or better ordered. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-2-review) - Name the framework you are reviewing against. Do not review in a vacuum. Pick one and be explicit: — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-2-review) - BLUF - does the bottom line appear in the first sentence? — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-2-review) - MECE - are sections mutually exclusive and collectively exhaustive? — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-2-review) - Minto Pyramid - does the argument flow deductively from a single governing idea? — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-2-review) - PARA - is the document in the right context (project, area, resource, or archive)? — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-2-review) - SCQA - is the Situation / Complication / Question / Answer arc intact? — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-2-review) - Severity-label every finding: Critical (document is wrong or misleading), High (missing material the audience needs), Medium (structure or argument weakness), Low (style preference, nice-to-have). — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-2-review) - Do not fix during review. Record findings. Mixing evaluation and correction corrupts both modes - you stop seeing what is actually there. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-2-review) ## Pass 3 — Revise - Mode: surgical. Apply the review findings against a priority threshold. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-3-revise) - Address every Critical and High finding. These are not optional. If a Critical finding cannot be resolved, the document is not ready to ship - restart or descope. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-3-revise) - Batch Medium findings. Group related Medium findings and address them in one pass through the affected section. Avoid thrashing individual sentences. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-3-revise) - Skip Low findings unless trivial. A Low finding that takes thirty seconds to fix is worth taking. A Low finding that requires restructuring a paragraph is not worth taking on the first revision cycle - bank it for the next iteration if it recurs. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-3-revise) - Do not introduce new content during revision. If revision uncovers a missing section, that section goes back to draft mode (Pass 1), not inline into the revision pass. Keep the modes clean. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-3-revise) - Log what changed and why. One-line entry per finding resolved: [H] §3 - added missing latency data; source: internal benchmark doc. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-3-revise) - See the Skill Handoff Map at the end of this document for routing after revision. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#pass-3-revise) ## Stop Conditions - The loop ends when the first applicable condition is met - earliest wins. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#stop-conditions) - If you reach the hard stop without reaching soft stop, ship the best available version with a known-issues note, or escalate the scope problem. Infinite iteration is not a quality strategy. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#stop-conditions) ## The 24-Hour Cooling-Off Rule - Stephen King recommends letting a draft rest before revising. The principle: you cannot read what you wrote - you read what you meant to write. Distance corrects that. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#the-24-hour-cooling-off-rule) - For documents under 500 words: 1–2 hours is sufficient. For documents 500–2,000 words: overnight. For documents over 2,000 words: 24 hours minimum. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#the-24-hour-cooling-off-rule) - If the deadline prevents cooling off, use the fresh eyes pattern instead. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#the-24-hour-cooling-off-rule) ## The Fresh Eyes Pattern - If you cannot wait, get a different reviewer for iteration N+1 than for iteration N. The first reviewer has already accommodated your argument; they will find it harder to see what is still missing. A second reviewer reads with no accommodation debt. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#the-fresh-eyes-pattern) - Applied to Claude: ask a second agent session (or invoke document-critique with a cold context) rather than asking the same session that helped draft to also review. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#the-fresh-eyes-pattern) ## Per-Iteration Log - Keep a running log for every document that goes through more than one cycle. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#per-iteration-log) ## Anti-Patterns - Editing while drafting. Kills generative momentum. You end up with the first three paragraphs polished and the rest unwritten. Lamott diagnosed this decades ago; the fix is the same: write badly first. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#anti-patterns) - Infinite polishing. Zinsser: "Rewriting is the essence of writing well - but there is a difference between rewriting that improves and rewriting that avoids." Apply the convergence and time-box stop conditions ruthlessly. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#anti-patterns) - Five-stakeholder review before any iteration. Gathering input from five people on a first draft means gathering five reactions to a document that will change dramatically anyway. Get to iteration 2 before widening the reviewer pool. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#anti-patterns) - No time-box on review. A review pass without a time-box expands to fill all available attention. Cap each review pass at half the draft time or 30 minutes, whichever is less. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#anti-patterns) - Skipping the framework. Reviewing without a named framework produces impressionistic feedback ("this feels off") that is hard to act on. Name the lens before you start. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#anti-patterns) ## When to Bail Entirely - Iteration is not always the answer. Stop the loop and start over when: — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#when-to-bail-entirely) - Scope changed. The document is now answering a different question than it started with. Revising toward a moving target produces incoherent documents. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#when-to-bail-entirely) - Audience changed. A document written for engineers and redirected to executives needs a new draft, not another revision pass. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#when-to-bail-entirely) - Premise was wrong. The core argument has been disproven or superseded. No amount of revision fixes a document whose central claim is no longer true. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#when-to-bail-entirely) - In all three cases: close the log, open a new one, and return to Pass 1. The old draft is not wasted - it is a source of raw material and a record of what the argument was. — [source](https://llms-explorer.com/sources/mdb-context-hub/draft-review-revise-loop/#when-to-bail-entirely) ## Where this helps - A long-form document (a spec, a customer-facing report, an RFC) needs to go from a rough argument to something ready to ship without the writer polishing sentences before the structure is settled. — [source](https://llms-explorer.com/tree/draft-review-revise-loop/) *(AI-suggested, synthesized from this pack's existing facts — not extracted from a source document.)* - A writer is prone to either shipping underdeveloped drafts or endlessly polishing without ever finishing — the loop's stop conditions catch both failure modes. — [source](https://llms-explorer.com/tree/draft-review-revise-loop/) *(AI-suggested, synthesized from this pack's existing facts — not extracted from a source document.)* - A piece of writing needs a genuine second read from a different vantage point (the Fresh Eyes pattern) rather than the same person editing their own prose in the same sitting. — [source](https://llms-explorer.com/tree/draft-review-revise-loop/) *(AI-suggested, synthesized from this pack's existing facts — not extracted from a source document.)* - A team wants a repeatable, auditable writing process with an explicit pass log, so reviewers can see which findings were addressed and which were deliberately deferred. — [source](https://llms-explorer.com/tree/draft-review-revise-loop/) *(AI-suggested, synthesized from this pack's existing facts — not extracted from a source document.)* ## How to apply this - Run Pass 1 as pure generation — write the shitty first draft, mark unknowns with [TK] or [GAP: ...] instead of stopping to research, and resist editing sentences mid-flow. — [source](https://llms-explorer.com/tree/draft-review-revise-loop/) *(AI-suggested, synthesized from this pack's existing facts — not extracted from a source document.)* - Run Pass 2 as a full read-through before any inline notes, then classify every finding by severity (Critical/High/Medium) so Pass 3 knows what's non-negotiable. — [source](https://llms-explorer.com/tree/draft-review-revise-loop/) *(AI-suggested, synthesized from this pack's existing facts — not extracted from a source document.)* - Apply the 24-hour cooling-off rule between drafting and reviewing when the schedule allows it, so the reviewer isn't still holding the writer's intent in working memory. — [source](https://llms-explorer.com/tree/draft-review-revise-loop/) *(AI-suggested, synthesized from this pack's existing facts — not extracted from a source document.)* - In Pass 3, resolve every Critical and High finding first, batch related Medium findings into one sweep through the affected section, and use a known stop condition to decide when to ship. — [source](https://llms-explorer.com/tree/draft-review-revise-loop/) *(AI-suggested, synthesized from this pack's existing facts — not extracted from a source document.)* ## Common mistakes - Editing sentence-level prose during Pass 1, which kills drafting momentum and produces a shorter, more cautious first draft than the ideas actually warrant. — [source](https://llms-explorer.com/tree/draft-review-revise-loop/) *(AI-suggested, synthesized from this pack's existing facts — not extracted from a source document.)* - Reviewing sentence-by-sentence instead of reading the whole draft once first — early inline notes get invalidated once the reviewer sees where the document actually goes. — [source](https://llms-explorer.com/tree/draft-review-revise-loop/) *(AI-suggested, synthesized from this pack's existing facts — not extracted from a source document.)* - Treating every Medium or Low finding as equally urgent in Pass 3, which causes thrashing on individual sentences instead of batched, efficient revision. — [source](https://llms-explorer.com/tree/draft-review-revise-loop/) *(AI-suggested, synthesized from this pack's existing facts — not extracted from a source document.)* - Iterating past the loop's hard stop condition in search of a perfect draft — the loop treats infinite iteration as a scope problem to escalate, not a quality strategy. — [source](https://llms-explorer.com/tree/draft-review-revise-loop/) *(AI-suggested, synthesized from this pack's existing facts — not extracted from a source document.)* ## Known issues - The 24-hour cooling-off rule assumes slack in the schedule; under real deadline pressure the loop has to compress passes, which weakens the Fresh Eyes benefit. — [source](https://llms-explorer.com/tree/draft-review-revise-loop/) *(AI-suggested, synthesized from this pack's existing facts — not extracted from a source document.)* - The loop is a coordination discipline, not a guarantee of quality — a reviewer with the same blind spots as the writer will still miss the same problems. — [source](https://llms-explorer.com/tree/draft-review-revise-loop/) *(AI-suggested, synthesized from this pack's existing facts — not extracted from a source document.)* - Marking gaps with [TK] defers research to Pass 3, which can surprise the timeline if a marked gap turns out to need substantial new investigation, not a quick fill. — [source](https://llms-explorer.com/tree/draft-review-revise-loop/) *(AI-suggested, synthesized from this pack's existing facts — not extracted from a source document.)* - The pass log adds real overhead for short, low-stakes writing; the loop is built for documents where the coordination cost is worth paying, not every email or Slack message. — [source](https://llms-explorer.com/tree/draft-review-revise-loop/) *(AI-suggested, synthesized from this pack's existing facts — not extracted from a source document.)* ## Context files - [Draft Review Revise Loop](https://llms-explorer.com/downloads/sources/mdb-context-hub/draft-review-revise-loop.md)