app-builder

A handoff guide for building a durable data app with a dashboard and background processes. It gives an app-building sub-agent the user's intended outcome and whether to use a new private workspace or an existing shared workspace.

In plain words
What is it for?
Use it when someone wants an app that stores or organizes data, displays a dashboard, and runs recurring or background work. It is not intended for a one-off widget or chart.
Why use it?
It keeps the main agent from having to design the app's data model, interface, and background processes itself. The user only needs to provide the app's purpose and workspace choice.

Skill for Claude CodeCodex

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/bearlike/assistant/app-builder
Any agent
npx skills add bearlike/Assistant --skill app-builder
Clone the repo
git clone --depth 1 https://github.com/bearlike/Assistant

Made for: Claude Code, Codex.

Per session 53 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 547 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.

ModelPer sessionOnce invoked
Fable 5 $0.00053 $0.00547
Opus 5 $0.00026 $0.00273
Sonnet 5 $0.00011 $0.00109
Haiku 4.5 $0.00005 $0.00055

Measured 2d ago against content hash b57eee86d033, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

app-builder 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 2d ago.

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.

apps/mewbo_api/src/mewbo_api/apps/plugin/skills/app-builder/SKILL.md · 56 lines

How it starts

The opening of the file, as written. The whole thing — 56 lines — stays where its author put it; the contents beside it link to each section on GitHub.

app-builder

app-builder is a sub-agent — delegate to it via spawn_agent. Do not design the data model, write the frontend, or author pipelines yourself.

How to invoke

spawn_agent(
  agent_type="app-builder",
  task="<intent + workspace choice — see below>",
  acceptance_criteria="app_ready event emitted",
)

What goes in the task

The sub-agent designs the collections, frontend, and pipelines itself. It handles both live/render apps and periodic/collection-backed apps. Worked pipeline shapes live at ${CLAUDE_PLUGIN_ROOT}/examples/recipes/.

Your task gives it only two things:

  1. Intent — what the user wants the app to do, in their own terms (what it tracks, who it's for, how often it should refresh).
  2. Workspace choice — which workspace the app's agents anchor to: own (a fresh private workspace for this app) or shared (an existing one, named). This is the one structural decision the user makes at creation; pass it through, don't invent it.
# Correct — intent + workspace, no layout
task=(
  "Build an app that organizes my inbox into actionable task groups every morning "
  "and shows them on a dashboard. Workspace: own."
)

# Wrong — prescribing the data model or the UI
task=(
  "Create two collections `emails` and `groups`, a cron pipeline, and a Streamlit "
  "page with a sidebar and three tabs..."
)

Do not prescribe collection schemas, the frontend layout, page structure, or pipeline mechanics. The sub-agent owns those decisions — prescribing them fights its judgment and produces a worse app.

After the sub-agent finishes

The app is live the moment app_ready fires. Reply with one sentence — "Done, your app is live." Don't describe what it shows; the user can open it.

What NOT to do

  • Do not write frontend code, collection schemas, or pipeline prompts yourself.
  • Do not call activate_skill("app-builder") — that does nothing; use spawn_agent.
  • Do not describe the app's layout or data model in the task — hand off intent and let the sub-agent design it.

Read the full file on GitHub · 56 lines

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.

  1. 2d ago First seen · 56 lines · 53 tokens per session scan A b57eee86d033

Subscribe to this mod's changes

app-builder is a skill published in the GitHub repository bearlike/Assistant (41 stars, last pushed 8d ago), licensed MIT. It adds 53 tokens to every session and 547 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.