Phase Build logo

Phase Build

Community
XuanRanL
phase-build

Run Phase Build — section drafting (N parallel agents), fact-check, citation injection, assembly. Use when research.json and outline.json exist and need to write the actual article body. Triggered by /seo-blog build, "draft the article now", "write the body".

Overview

PublisherXuanRanL
Repositoryloamwright-SEO-Skill
Skill namephase-build
Stars
52
Forks
9
Bundled files
Instructions only
LicenseApache-2.0
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by XuanRanL on GitHub. Read the source before you install it.

Installation

Install the Phase Build AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/XuanRanL/loamwright-SEO-Skill/tree/main/skills/phase-build
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/XuanRanL/loamwright-SEO-Skill.git /tmp/loamwright-SEO-Skill
mkdir -p .claude/skills
cp -r /tmp/loamwright-SEO-Skill/skills/phase-build .claude/skills/phase-build
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Phase Build in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Phase Build on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Phase Build is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Phase Build Orchestrator

Convert outline.json → full markdown draft via parallel section writers + fact verification.

Inputs

Required in workspace/{task_id}/:

  • state.json
  • research.json
  • angle.json (from format-selector + topic-angle-selector)
  • outline.json (from outline-architect)

Stages

1. section-drafter (parallel N=outline.sections.length)
   - Spawn N independent subagents via Task tool
   - Each gets: ONE section spec + 800-word context summary of others
   - Each writes markdown + claims[] + citation_capsule
   - Tool whitelist: Read, Write ONLY (no WebFetch, no Bash)
   - Output: workspace/{task}/sections/{0..N-1}.json
   
2. citation-capsule-builder
   - Verify each H2 has 40-60 word AI-quotable block
   - Princeton GEO requirement (+28-41% AI citation rate)

3. chart-generator (if outline.sections[].needs_table)
   - Generate ≥2 markdown tables OR XSS-safe SVG charts
   - At least 1 in front 50% of article

4. fact-check-and-citation
   - For each [claim:cN_S] marker: Crossref → Tavily → HEAD check
   - Build References section (APA 7, ≤10 entries)
   - Replace markers with (Author, Year) in-text
   - Output: workspace/{task}/citations.json

5. assembly (scripts/build/assemble.py)
   - Concatenate: Abstract → Key Takeaways → ToC → Body → FAQ → Conclusion → References
   - Anchor links auto-generated for ToC
   - Mandatory-H2 coverage is enforced by scripts/lint/mandatory_sections_check.py,
     which reads THIS project's `business-context.json :: mandatory_sections` (assemble.py
     imports `_load_mandatory_sections` from it). Do NOT substitute
     markdown_structure_check.py here: it carries a HARDCODED section table and would
     override the project contract (see feedback_project_contract_not_hardcoded_labels).
     It remains available as a standalone structural report, not as this stage's gate.
   - Output: workspace/{task}/draft.md (frontmatter Stage: build-done)

Parallel execution pattern

Section drafter spawns subagents in single Task batch:

python
tasks = []
for section in outline.sections:
    tasks.append({
        "subagent_type": "writer",
        "section_spec": section,
        "context_summary": summarize_others(outline, section),
        "primary_keyword": state.brief.primary_keyword,
        "format_id": angle.format_id,
        "modifiers": angle.modifiers,
        "references_to_load": [
            "style/voices/" + brand_voice_choice,
            "style/banned-words.md",
            "style/ai-tells-43.md",
            "style/em-dash-prohibition.md",
            "seo/citation-capsules-princeton.md"
        ]
    })

results = parallel_dispatch(tasks)  # 1 round-trip, N writers

Hard constraints (enforced post-section)

Each section must pass self_check:

  • word_count within [budget × 0.9, budget × 1.1]
  • em_dash_count == 0
  • banned_words_found == false
  • ai_tells_found == false
  • has citation_capsule of 40-60 words
  • contains claims with hint_query for fact-checker

Otherwise section is rejected and drafter retried (max 2 retries per section).

Post-build lint (mandatory before handoff to Optimize)

bash
# Section completeness — catches writer subagent silent dropout (6/50 sections
# dropped in 2026-05-22 batch). Run immediately after Stage 1 parallel dispatch
# returns, BEFORE Stage 4 fact-check.
python -m scripts.lint.section_completeness_check --workspace {task_id} --json
# exit 0 = all sections present; exit 1 = missing or extra
# Auto-writes to workspace/{task_id}/section-completeness.json (file-bus)
# Missing indices → re-dispatch writer subagent for each before proceeding.

Output

  • workspace/{task}/sections/*.json (N files, schemas/section.schema.json)
  • workspace/{task}/citations.json (schemas/citations.schema.json)
  • workspace/{task}/draft.md (markdown with frontmatter Stage: build-done)

Handoff

recommended_next_skill: "phase-optimize"

See also

  • subskills/build/section-drafter/SKILL.md
  • subskills/build/fact-check-and-citation/SKILL.md
  • subskills/build/citation-capsule-builder/SKILL.md
  • agents/writer.md
  • agents/fact-checker.md

Frequently asked questions

What does the Phase Build AI skill do?

Run Phase Build — section drafting (N parallel agents), fact-check, citation injection, assembly. Use when research.json and outline.json exist and need to write the actual article body. Triggered by /seo-blog build, "draft the article now", "write the body".

Why use Phase Build on TypingMind?

Because you install it once and use it with any model. Phase Build is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Phase Build in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/XuanRanL/loamwright-SEO-Skill/tree/main/skills/phase-build. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Phase Build?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Phase Build?

As many as you like. As long as a model supports skills, you can use Phase Build with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Phase Build AI skill free?

Yes. It is published on GitHub by XuanRanL under the Apache-2.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇