Skills for making web pages that are contextually relevant to your audience and what your product does. Ships the 7-step method (signal doc, message spine, first-page decision, page argument shape, proof map, first-page draft, claim check) plus tooling skills and subagents.
npx claudepluginhub hnshah/pagekit --plugin pagekitSkeptical, read-only claim-check pass on a PageKit first-page draft. Runs at the requested severity (default hard). Produces an audit (claim-check.md) and a corrected draft (first-page-draft-corrected.md). Use when the user asks for a claim-check on a draft, or when the pagekit-claim-check skill delegates here.
Skeptical, read-only adversarial second read of a completed PageKit run. Produces runs/<name>/evaluator-pass.md. Runs after evaluation.md exists. Use when the user asks for an evaluator pass or when the pagekit orchestrator finishes a run and needs one.
Run a skeptical claim-check pass on a PageKit first-page draft, at light/normal/hard severity. Step 07 of the PageKit method. Use after a first-page draft exists and you want to validate it before treating it as publishable. Delegates to the pagekit-claim-checker subagent.
Run the adversarial evaluator pass on a completed PageKit run. Post-evaluation skeptical second read. Use when a run has an evaluation.md but no evaluator-pass.md, or when the orchestrator needs an evaluator pass. Delegates to the pagekit-evaluator-pass subagent.
Decide which page should exist first for the object, without defaulting to homepage. Step 03 of the PageKit method. Use when the user has a signal doc + message spine and needs to choose the first page, or asks "should this be a homepage?" or "what page should we build first?"
Draft the first page from the full PageKit chain. Step 06 of the PageKit method. Use when steps 01-05 are done and you are ready to write the page itself. The draft must respect the anti-slop drafting constraints.
Reduce a PageKit signal doc into a message spine. Step 02 of the PageKit method. Use when the user asks to "reduce this signal to a spine," "build a message spine," or has finished step 01 and is ready for step 02.
Scaffold a new fully-logged PageKit run folder by name. Wraps scripts/new-run.sh. Use when starting any PageKit run; prefer this over creating run folders by hand so structure stays consistent with the run-logging framework.
Design the page argument shape — what sections the chosen first page should and should not have. Step 04 of the PageKit method. Use when the first-page decision is made and you need a structure for the draft.
Map proof to the chosen first page section by section. Step 05 of the PageKit method. Use after the page argument shape exists and before drafting.
Validate a PageKit run folder against the logging tiers (fully-logged / summary-logged / artifact-only / incomplete). Wraps scripts/run-check.sh. Use at the end of a run to confirm the fully-logged tier is met before declaring done.
Build a PageKit signal doc from a product description and any source briefs. Step 01 of the PageKit method. Use when the user asks to "build a signal doc," "extract signal from this material," or is starting a PageKit run on a new object.
Run the heuristic AI-slop regression check on PageKit drafts. Wraps scripts/slop-check.sh. Use before and after the claim-check step, and any time you want to sanity-check a draft for the mechanically detectable slop patterns.
Run the full PageKit method end-to-end on an object. Use when the user asks to "run PageKit on X", "build a page for X using PageKit", or "produce a fully-logged PageKit run for X". Orchestrates steps 01-07 (signal doc, message spine, first-page decision, page argument shape, proof map, first-page draft, optional claim-check) and produces a fully-logged run on disk under runs/<name>/.
Skills for making web pages that are contextually relevant to your audience and what your product does.
PageKit ships as a Claude Code plugin. One install, fourteen skills (twelve method skills + two subagents), and the full 7-step chain runs end-to-end: signal doc → message spine → first-page decision → page argument shape → proof map → first-page draft → claim check. An anti-slop regression script catches mechanical slop patterns; the claim-check subagent catches the rest.
If you do not use Claude Code, there is a chat-paste path in prompt-paste/.
Inside Claude Code:
/plugin marketplace add hnshah/pagekit
/plugin install pagekit@pagekit
That is the whole installation. All fourteen skills register automatically.
Ask Claude Code to run PageKit on any object:
Run PageKit on <product or idea>.
The orchestrator skill (pagekit) walks you through the seven steps, scaffolds a fully-logged runs/<name>/ folder, and invokes each step skill in turn.
To run one step on its own, invoke the step skill directly:
Use pagekit-signal-doc on <product>.
| Skill | What it does |
|---|---|
pagekit | Orchestrator. Runs all 7 steps end-to-end into a fully-logged run. |
pagekit-signal-doc | Step 01. Turn source briefs into a signal doc. |
pagekit-message-spine | Step 02. Reduce the signal doc to a message spine. |
pagekit-first-page-decision | Step 03. Decide which page comes first — without defaulting to homepage. |
pagekit-page-argument-shape | Step 04. Shape the argument for the chosen page. |
pagekit-proof-map | Step 05. Map proof to sections. Name what cannot be said. |
pagekit-first-page-draft | Step 06. Draft the page inside the argument shape and proof map. |
pagekit-claim-check | Step 07. Adversarial audit at light / normal / hard severity. |
pagekit-new-run | Scaffold a fully-logged run folder. |
pagekit-run-check | Validate a run against the logging tiers. |
pagekit-slop-check | Heuristic anti-slop regression against drafts. |
pagekit-evaluator-pass | Invoke the evaluator-pass subagent on a completed run. |
Two subagents (pagekit-claim-checker, pagekit-evaluator-pass) do the adversarial reads. The orchestrator dispatches to them automatically.
examples/vegan-dog-food-verdel/ is the canonical run. Fully logged, non-homepage first page, claim-checked at severity: hard, evaluator pass included. Read first-page-draft.md against proof-map.md against claim-check.md to see how the chain constrains the draft.
examples/personal-crm-founders/ is a second worked example on a different object shape.
If you use ChatGPT, Perplexity, Grok, Gemini, or another chat window, open prompt-paste/ and paste the 7 prompts in order. Your output quality depends on how honest your source briefs are; that constraint does not change.
pagekit-slop-check runs the regression; pagekit-claim-check catches the rest..claude-plugin/: plugin and marketplace manifestsskills/: 12 SKILL.md files, each with a bundled references/ folderagents/: 2 subagents (claim-checker, evaluator-pass)examples/: worked runs (vegan-dog-food-verdel, personal-crm-founders)prompt-paste/: chat-app path with 7 prompts ready to pasteruns/: your own runs land here (empty by default)scripts/: tooling (new-run, run-check, claim-check, slop-check, doctor)MIT
Built by Hiten.
Upstash Context7 MCP server for up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your LLM context.
Uses power tools
Uses Bash, Write, or Edit tools
Comprehensive startup business analysis with market sizing (TAM/SAM/SOM), financial modeling, team planning, and strategic research
Semantic search for Claude Code conversations. Remember past discussions, decisions, and patterns.