agentsclimarketplace
Sign in

Doc writing

Skill addxai/enterprise-harness-engineering/skills/doc-writing

Write documents using the HWPR/AWOR framework -- separating human value judgments from AI-expanded content so critical information is not buried. Triggers when the user requests writing, rewriting, or reviewing document quality.From its SKILL.md

Install
npx -y skills add addxai/enterprise-harness-engineering --skill doc-writing

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

SKILL.md

5.8 KB, ~1.3k tokens by cl100k_base, as published. Nobody here has run it

Doc Writing (HWPR/AWOR)

In the AI era, human value judgments get buried in AI-expanded long documents. This skill uses HWPR/AWOR markers so readers (human or AI) can quickly locate what the human actually thought.

For the detailed template, see examples/TEMPLATE-HWPR.md.

Core Concepts

  • [HWPR] (Human Wrote, Please Read): Unknown context + value judgments written by a human. Must be short (3-5 sentences).
  • [AWOR] (AI Wrote, Optional Read): Detailed content expanded by AI. Can be deleted, modified, or replaced.

Rules

  1. Never modify [HWPR] content — AI may only read HWPR paragraphs; it must not rewrite, rephrase, merge, or "polish" them
  2. HWPR must be short — Each HWPR paragraph must not exceed 3-5 sentences; write only: unknown context + value judgments
  3. Value judgments with humility — Use phrasing like "I believe" / "current judgment" / "possibly" in HWPR, acknowledging potential error
  4. [AWOR] can be freely modified — AI-expanded content may be replaced, deleted, or rewritten at any time
  5. Consistent marker format — Use bold markers **[HWPR]** and **[AWOR]** as headers, followed by paragraph titles
  6. HWPR uses blockquote — HWPR body text uses > block quotes for visual distinction

Execution Flow

Mode A: Write a New Document

Trigger Conditions

User requests "help me write a document," "write a proposal," "draft a PRD," etc.

Step 1: Guide HWPR Extraction

Ask the user questions to extract core value judgments:

To write an effective document, I need you to provide the following HWPR content (keep it brief, 1-3 sentences per item):

1. **Background**: Why are we doing this? What is the core problem?
2. **Judgment**: What do you think we should do? Why this direction?
3. **Trade-offs**: What was deliberately given up? What are the known risks?

Step 2: Confirm HWPR

Organize the user's answers into HWPR paragraphs and display them for user confirmation. Once confirmed, HWPR is never modified afterwards.

Step 3: Generate Complete Document

Following the TEMPLATE-HWPR.md structure, expand corresponding AWOR paragraphs after each HWPR paragraph.

Mode B: Rewrite an Existing Document

Trigger Conditions

User provides an existing document and requests "restructure using HWPR/AWOR," "split and label," etc.

Step 1: Identify Potential HWPR

Read the full text and mark sentences/paragraphs that appear to contain human value judgments (identification criteria: contains subjective decisions, trade-offs, "we chose" / "gave up" language, etc.).

Step 2: Confirm with User

List the identified results and ask the user to confirm each one:

I identified the following as potentially your value judgments (HWPR) in the document. Please confirm:

1. yes/no "We chose option B because..." (paragraph X)
2. yes/no "Abandoned real-time push, switched to polling..." (paragraph Y)
3. yes/no ...

Step 3: Split, Label + Expand

Extract confirmed HWPR into **[HWPR]** paragraphs, mark remaining content as **[AWOR]**, and expand where necessary.

Mode C: Review a Document

Trigger Conditions

User requests "review the document," "check HWPR formatting," etc.

Review Checklist

Check and report the following issues:

Check ItemIssue Description
Missing markersParagraph has no [HWPR] or [AWOR] marker
HWPR too longHWPR paragraph exceeds 5 sentences
HWPR contains AI styleHWPR has obvious AI-expansion artifacts (boilerplate, "in summary," etc.)
AWOR contains value judgmentsAWOR contains "we decided" / "gave up" etc. that should be HWPR content
Incorrect marker formatNot using the standard **[HWPR]** / **[AWOR]** format

Output format: List each issue + suggested fix.

Examples

Bad — HWPR too long, mixed with AI style

**[HWPR]** Background and Judgment
> After in-depth analysis of user behavior data and multi-dimensional competitive market research,
> our team discovered that the core problem lies in the new user onboarding experience not being smooth enough,
> which has led to a first-day retention rate of only 35%, significantly below the industry average of 50%.
> Based on the above analysis, we believe we should start by simplifying the onboarding flow,
> improving user experience through reducing step count and optimizing interaction design... (200 words)

Problem: HWPR is too long; contains AI boilerplate ("after in-depth analysis," "multi-dimensional," "significantly below").

Good — HWPR is concise, focused on judgments

**[HWPR]** Background
> New user first-day retention is 35%. I believe the main cause is onboarding being too complex (5 steps).
> Plan to simplify to 2 steps first, targeting 45% retention.

**[AWOR]** Detailed Analysis
User growth data over the past three quarters: Q1 retention 38%, Q2 35%, Q3 33%, showing a continuous decline.
Competitor comparison: Product A's onboarding has only 2 steps with 52% first-day retention...

Exemptions

ScenarioCondition
Pure record documentsMeeting minutes and other pure records without value judgments — HWPR may be omitted
Existing mature templatesWeekly reports and other documents with fixed formats — only add HWPR to "judgment/decision" sections

References

What ships with it: 1 file

963 B alongside SKILL.md

examples/

Gives 0 of the 12 instructions most docs writing skills give in ~1.3k tokens

Counted across 1,951 of the 3,904 authors here whose files we hold, read 2026-09-06

  • Use third-person for skill descriptionsin 54 of 1951, across 35 files
  • Start descriptions with Use whenin 43 of 1951, across 29 files
  • Run baseline scenarios before writing any skillin 40 of 1951, across 26 files
  • Use active voicein 40 of 1951, across 36 files
  • Map file responsibilities before defining tasksin 36 of 1951, across 29 files
  • Use checkbox syntax for tracking stepsin 35 of 1951, across 27 files
  • Ask one question at a timein 35 of 1951
  • Offer execution options after saving the planin 33 of 1951, across 24 files
  • Include complete code in every stepin 33 of 1951, across 27 files
  • Design units with clear boundaries and interfacesin 31 of 1951, across 23 files
  • Announce the skill usage at the startin 30 of 1951
  • Verify agent compliance after adding the skillin 29 of 1951, across 17 files

Said here and by no other author read

  • Use bold markers for HWPR and AWOR headers
  • Use block quotes for HWPR body text
  • Ask user questions to extract core value judgments
  • Display HWPR content for user confirmation before expansion
  • Expand AWOR content after each HWPR paragraph
  • Identify human value judgments in existing documents

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.