An iterative website-building workflow for Stitch, a tool that generates interface screens from descriptions. It uses a file called next-prompt.md as a handoff note between iterations, building and integrating one page at a time.

Use it to generate a site page by page, integrate each result, and continue the build across repeated Stitch iterations.

It keeps multi-page generation organized and helps later pages follow the same design system. The handoff file records what should be built next.

stitch-loop — Skill by gabelul

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods

npx agentmods add skills/gabelul/stitch-kit/stitch-loop

Any agent

npx skills add gabelul/stitch-kit --skill stitch-loop

Clone the repo

git clone --depth 1 https://github.com/gabelul/stitch-kit

Made for: Claude Code, Codex.

Per session 65 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 1,567 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. Scan, not verified.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5	$0.00065	$0.01567
Opus 5	$0.00032	$0.00783
Sonnet 5	$0.00013	$0.00313
Haiku 4.5	$0.00006	$0.00157

Measured yesterday against content hash 64b9b510f578, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

stitch-loop scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/stitch-loop/SKILL.md · 184 lines

How it starts

The opening of the file, as written. The whole thing — 184 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Stitch Build Loop

Constraint: Only use this skill when the user explicitly mentions "Stitch" and multi-page or iterative site building.

You are an autonomous frontend builder in an iterative site-building loop. Each iteration: (1) Read the baton, (2) Generate a page with Stitch MCP, (3) Integrate into the site, (4) Write the next baton so the loop continues.

Prerequisites

Stitch MCP Server (see stitch-setup skill or https://stitch.withgoogle.com/docs/mcp/guide/)
DESIGN.md — generate with stitch-design-system from an existing screen (required for consistency)
SITE.md — site vision, Stitch project ID, sitemap, roadmap (create if missing)

The baton system

next-prompt.md is the relay baton between iterations. It tells the loop what page to build next.

Baton format

---
page: about
---

**DESIGN SYSTEM (REQUIRED):**
[Paste DESIGN.md Section 6 here verbatim]

**Page request:**
About page with company mission, team section (3 people), and timeline.

Rules:

page frontmatter → output filename (about.html)
The body must include the design system block from DESIGN.md Section 6
You must update next-prompt.md at the end of every iteration — the loop stops if this is missing

Execution protocol

Step 1 — Read the baton

Parse next-prompt.md:

Extract page from YAML frontmatter
Extract the full prompt body (including the design system block)

Step 2 — Consult context files

File	What to look for
`SITE.md`	Stitch project ID (Section 2), sitemap (Section 3), roadmap / next pages (Section 4), creative freedom (Section 5)
`DESIGN.md`	Section 6 — copy this into every prompt, not just the current one

Check: Do not rebuild pages already in SITE.md sitemap. Do not deviate from the visual language in DESIGN.md.

Step 3 — Generate with Stitch

Run list_tools → find Stitch MCP prefix
If stitch.json exists, use stored projectId — do not create a new project
If no stitch.json, call create_project → save numeric ID to stitch.json
Call generate_screen_from_text:
- projectId: numeric ID (no projects/ prefix)
- prompt: full baton content including DESIGN SYSTEM block
- deviceType: match what's in SITE.md or DESIGN.md
Call get_screen with numeric projectId + screenId
Download HTML: bash scripts/fetch-stitch.sh "[htmlCode.downloadUrl]" "queue/[page].html"
Download screenshot: save to queue/[page].png

Read the full file on GitHub · 184 lines

Files

What ships with it

3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

yesterday First seen · 184 lines · 65 tokens per session scan A 64b9b510f578

Subscribe to this mod's changes

stitch-loop is a skill published in the GitHub repository gabelul/stitch-kit (43 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 65 tokens to every session and 1,567 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Data and AI Frontend Mobile claude-code claude-code-plugin design-to-code nextjs react-native stitch

karakeep

Official skill for how to use karakeep (the bookmark manager) and interact with it programmatically.

karakeep-app/karakeep · 25 tokens

revenuecat

Integrate RevenueCat for in-app purchases — consumable credit packs or subscriptions. Use when setting up RevenueCat, creating products in App Store Connect / Google Play, configuring the SDK, building a paywall, or handling purchase webhooks in Supabase Edge Functions.

Ampli-Group/agentic-mobile-blueprint · 57 tokens

template-upgrade

Upgrade a project that is based on the agentic-mobile-blueprint template. Use when the user asks to check for template updates, upgrade to a newer version, or apply template changes.

Ampli-Group/agentic-mobile-blueprint · 41 tokens

mobile-development

React Native mobile app development patterns with Expo 54, Tamagui, and expo-router. Use when creating screens, components, or navigation in the mobile app.

Ampli-Group/agentic-mobile-blueprint · 36 tokens

sentry-integration

Add Sentry error tracking to Supabase Edge Functions. Use when creating functions that need error monitoring.

Ampli-Group/agentic-mobile-blueprint · 25 tokens

tamagui

Universal React UI framework for web and native. Use when building cross-platform apps with Tamagui, creating styled components with styled(), configuring design tokens/themes, using Tamagui UI components, or working with animations. Triggers: "tamagui", "styled()", "$token", "XStack/YStack", "useTheme", "@tamagui/"…

tamagui/tamagui · 90 tokens