Open-source design agent: on-brand stills, decks, diagrams, charts and motion graphics from a chat thread, built on Deep Agents.
JavaScript agentagentsai-agentdeepagentsdesignhyperframeslanggraphmanaged-agentsmotion-graphicspaperslackvideo
README: fix images in dark mode, conversational onboarding-first rewrite
Amal Irgashev73f47877aec8
diff --git a/README.md b/README.mdindex b6ec7e5..2e82c85 100644--- a/README.md+++ b/README.md@@ -2,13 +2,13 @@<p><a href="https://www.langchain.com/"><picture>- <source media="(prefers-color-scheme: dark)" srcset="docs/assets/readme/logos/langchain-company-dark.svg">- <img src="docs/assets/readme/logos/langchain-company-light.svg" alt="LangChain" width="152">+ <source media="(prefers-color-scheme: dark)" srcset="https://github.com/langchain-ai/design-agent/raw/main/docs/assets/readme/logos/langchain-company-dark.svg">+ <img src="https://github.com/langchain-ai/design-agent/raw/main/docs/assets/readme/logos/langchain-company-light.svg" alt="LangChain" width="152"></picture></a></p><h1>Design Agent</h1>- <p><strong>An open-source AI design agent</strong> that turns a chat thread into on-brand stills, decks,<br>diagrams, charts and motion graphics, built and checked on its own computer.</p>+ <p><strong>An open-source AI design agent.</strong> Ask for a launch card, a deck, a diagram, a chart or a<br>short film in Slack, and get finished, on-brand files back in the thread.</p><p><a href="https://github.com/langchain-ai/design-agent/actions/workflows/ci.yml"><img src="https://github.com/langchain-ai/design-agent/actions/workflows/ci.yml/badge.svg" alt="CI"></a><a href="LICENSE"><img src="https://img.shields.io/badge/license-Apache--2.0-blue" alt="License: Apache 2.0"></a>@@ -19,161 +19,60 @@<a href="https://paper.design"><img src="https://img.shields.io/badge/canvas-Paper-111111" alt="Canvas in Paper"></a></p><p>- <a href="#quick-start">Quick start</a> ·- <a href="#architecture">Architecture</a> ·- <a href="#deployment">Deployment</a> ·- <a href="OPERATIONS.md">Documentation</a> ·+ <a href="#get-started">Get started</a> ·+ <a href="#how-it-works">How it works</a> ·+ <a href="#deploy-it">Deploy</a> ·+ <a href="OPERATIONS.md">Docs</a> ·<a href="CONTRIBUTING.md">Contributing</a></p></div><picture>- <source media="(prefers-color-scheme: dark)" srcset="docs/assets/readme/hero-dark.png">- <img src="docs/assets/readme/hero-light.png" alt="Design Agent at work: a Slack request for a launch film, and the agent's computer with its project folders, a HyperFrames render in the terminal, the film preview, layers and references" width="100%">+ <source media="(prefers-color-scheme: dark)" srcset="https://github.com/langchain-ai/design-agent/raw/main/docs/assets/readme/hero-dark.png">+ <img src="https://github.com/langchain-ai/design-agent/raw/main/docs/assets/readme/hero-light.png" alt="Design Agent at work: a Slack request for a launch film, and the agent's computer with its project folders, a HyperFrames render in the terminal, the film preview, layers and references" width="100%"></picture>-Ask in Slack or the terminal for an ad set, a LinkedIn card, a deck, a diagram, a chart, a thumbnail-set, a storyboard or a short film. Design Agent makes the concept, layout and copy calls, builds the-work as HTML, SVG and video on its own sandboxed computer, checks it against your brand with-deterministic gates, and delivers the files to the thread. It is a-[Deep Agents](https://github.com/langchain-ai/deepagents) application with one shared core and two-deployment paths: [Managed Deep Agents](https://docs.langchain.com/langsmith/python/managed-deep-agents-overview)-in one command, or self-hosted with Docker.+We built Design Agent as our own design teammate at LangChain, and now it's open source.++You ask for something the way you'd ask a designer: "a LinkedIn card for Thursday's launch", "turn+this into a 2x2", "a 30-second teaser for the new release". It picks a concept, lays the piece out,+builds it on its own computer, checks it against your brand, and posts the files back to you. For+anything big, like a film, it shows you a storyboard first, and you approve it before it renders.> [!NOTE]-> The images and film in this README use LangChain's brand, the brand we run it with. Design Agent-> ships with an original sample brand, and every output follows the brand pack you give it: build-> yours with the [brand-onboarding skill](.agents/skills/brand-onboarding/SKILL.md).--**Contents:** [What it makes](#what-it-makes) · [How it works](#how-it-works) ·-[Architecture](#architecture) · [Quick start](#quick-start) ·-[Your brand, media and Paper](#your-brand-media-and-paper) · [Deployment](#deployment) ·-[FAQ](#faq) · [Built with](#built-with)--## What it makes--- **Stills and sets:** layouts rendered in a real browser, each size recomposed rather than scaled;- PDFs and decks; image edits; GIFs.-- **Diagrams and charts:** JSON specs rendered with ECharts and D2 on your brand, with the numbers- and the topology checked in code.-- **Motion:** storyboards first, then [HyperFrames](https://hyperframes.heygen.com) compositions- rendered to video, with voice, music and sound when keys are configured. Long films run as a- pipeline of background helpers that check in with you at the right moments.-- **On brand:** a brand pack (tokens, colour roles, logos, fonts, voice, graphic language) is the- agent's only source of brand facts, and gates check every output against it.-- **Paper:** layouts and storyboards go onto the [Paper](https://paper.design) canvas as editable- layers when Paper is connected.+> Everything you see here uses LangChain's brand, because that's the brand we run it with. Out of+> the box Design Agent uses a neutral sample brand, and it switches to yours as soon as you add+> your brand pack (see [Make it yours](#make-it-yours)).-## How it works+## See it work<p align="center">- <a href="docs/assets/readme/design-agent-explainer.mp4"><img src="docs/assets/readme/explainer-preview.gif" alt="From the explainer film: the lead hands out a film in three acts, and a flock of helper parrots works the act folders in parallel" width="100%"></a>- <br><sub><b><a href="docs/assets/readme/design-agent-explainer.mp4">Watch the 2-minute film</a></b>: a Slack request moving through the sandbox, skills, Paper, the helpers, HyperFrames, the checks and the sound.</sub>+ <a href="https://github.com/langchain-ai/design-agent/raw/main/docs/assets/readme/design-agent-explainer.mp4"><img src="https://github.com/langchain-ai/design-agent/raw/main/docs/assets/readme/explainer-preview.gif" alt="From the explainer film: the lead hands out a film in three acts, and a flock of helper parrots works the act folders in parallel" width="100%"></a>+ <br><sub><b><a href="https://github.com/langchain-ai/design-agent/raw/main/docs/assets/readme/design-agent-explainer.mp4">▶ Watch the 2-minute film</a></b>: one Slack request, from the sandbox to skills, Paper, the helpers, HyperFrames, the checks and the sound.</sub></p>-Three layers are specific to design work. **The lead and its helpers** plan the job and hand out the-pieces. **The context** is skills, a design wiki, the brand pack and memory, loaded only when a job-needs them. **The designer's computer** is a sandbox with real tools, where everything is built and-checked. Underneath, the agent is off the shelf: the Deep Agents harness, Managed Deep Agents (or-your own infrastructure) and a frontier model. The model makes the creative calls. Code renders,-measures, runs the gates, delivers files and holds the credentials.--## Architecture--One assembly (`src/design_agent/assembly.py`) builds the model, tools, middleware and helpers.-Runtime profiles plug it into each place it runs: Managed Deep Agents, the CLI, the self-hosted API,-Slack and schedules.--### The lead and its helpers--- **One lead per thread.** It reads the brief, decides the concept and writes a plan in the- thread's project folder. It does small jobs itself.-- **Helpers in the background.** Bigger jobs go to helpers that work in parallel on the same- sandbox and files: a storyboard, one act of a film, three directions to compare. Each helper has- a heartbeat and a durable transcript, and the lead can steer it.-- **Wake-ups instead of waiting.** Render watchers and check-ins wake the lead when work lands,- fails or stalls. On Managed Deep Agents these are one-time schedules on the thread; elsewhere,- an in-process waker persisted in the store.-- **Looking before delivering.** The lead and helpers review their own renders with a vision- `look` tool before any file is delivered.--### The context--- **A small resident prompt** of about 4k tokens. Workflows live in [skills](skills/)- (`skills/<name>/SKILL.md`), loaded on demand; on MDA they mount read-only from Context Hub.-- **Brand facts only from the brand pack**, staged into the sandbox with its fonts and logos. A- design wiki holds taste and doctrine, and memory holds what your team taught it.-- **Bounded context:**- - Long tool results go to files, with compact summaries in the conversation.- - Provider tool search loads MCP tools only when a job needs them.- - Helpers can clear old tool results server-side (`DESIGN_AGENT_ENABLE_TOOL_CLEARING`).- - Prompt caching keeps the lead's prefix warm for an hour.--### The designer's computer--- **One recipe for every sandbox.** [`sandbox/setup.sh`](sandbox/) pins Chrome Headless Shell,- FFmpeg, HyperFrames, sharp, ECharts, D2 and open fonts by version and sha256. It ends with a probe- that renders a page, a film and a diagram. The same recipe builds:- - the local and self-hosted Docker image;- - the LangSmith snapshot that Managed Deep Agents clones for each thread;- - the CI image.-- **One sandbox per thread.** The thread's files persist in `/workspace/projects/` between turns.- - **Locally:** a container with no network, dropped capabilities, an unprivileged agent user,- and limits capped to the host.- - **On MDA:** a LangSmith sandbox (16 GiB disk, 8 vCPU) with default-deny egress.-- **Staged, read-only inputs.** Before a run, a manifest with a sha256 checklist stages skills, the- gates, the brand pack and its fonts into `/workspace`. Licensed fonts arrive as deployment files- or a mount, never through Git.-- **Gates before delivery** ([`workspace/tools/`](workspace/tools/README.md)):-- | Gate | Checks |- | --- | --- |- | `brand-lint` | palette, fonts, radii and copy rules from the brand pack |- | `video-gate`, `hyperframes check` | determinism (no clocks or randomness), layout, motion and contrast |- | `artifact-contract` | every file exists at its size and duration, and frame 0 is not blank |- | font gate | every brand family actually paints, web fonts included |- | chart and graph contracts | zero baselines, one unit per axis, shares that sum to 100%, frozen topology |--### Keys stay outside--The sandbox never receives a credential. Host tools run beside the agent with the keys and drop-their results into the sandbox as files:--- image generation;-- voice, music and sound effects;-- transcription;-- an SSRF-hardened asset fetcher;-- Paper, scoped by a policy;-- your MCP servers;-- file delivery: to Slack, to `workspace/out/` locally, or attached to the LangSmith trace for- API runs.--### Where it runs+## What you can ask for-| | Managed Deep Agents | Self-hosted |-| --- | --- | --- |-| Agent | `agent.py` (`define_deep_agent`), `identity.py`, `tools/mcp.py` | the same assembly in a FastAPI app |-| Threads and memory | managed by LangSmith | Postgres checkpointer and store |-| Sandboxes | LangSmith sandboxes cloned from your snapshot (`sandbox/__init__.py`) | containers on your Docker host |-| Slack | managed app (`channels/slack.py`) | your Bolt app, Socket Mode or signed HTTP |-| Wake-ups and housekeeping | runtime schedules; `schedules/housekeeping.py` every 15 minutes | in-process waker and maintenance loop |-| Skills | synced to Context Hub; binary files ship as deployment files | read from the checkout |-| Deploy | `design-agent deploy` stages a minimal project and a filtered `.env` | `docker compose up` |--**Models.** Claude Opus 5.5 is the default, and any tool-calling model with vision works. A-capability registry keyed by exact model id sets effort, tool search and output limits for each-role: lead, helper, look, tool selector and learning. The optional-[LangSmith Gateway](docs/models.md) uses each provider's native route, so caching and context-editing keep working. Every call is metered per job, helper and role at list prices, and LangSmith-traces every run.+- **Social cards and ad sets:** one layout, recomposed for every size you need. It doesn't just+ scale it down.+- **Slides, one-pagers and decks**, as PNG or PDF.+- **Diagrams and charts**, drawn from your data. The numbers are computed in code, never+ eyeballed.+- **Thumbnails, banners and GIFs.**+- **Short films and motion graphics:** storyboard first, then a+ [HyperFrames](https://hyperframes.heygen.com) render, with voice, music and sound if you add+ the keys.+- **Work on a shared canvas:** layouts and storyboards land in [Paper](https://paper.design) as+ editable layers, if you connect it.-More in [Architecture](docs/architecture/README.md) ·-[Orchestration](docs/architecture/orchestration.md) ·-[Context and failures](docs/architecture/context-and-failures.md).+## Get started-## Quick start+You need **Python 3.11+**, [uv](https://docs.astral.sh/uv/) and **Docker**. Setup takes about five+minutes, plus a few more the first time Docker builds the agent's computer.-Requires **Python 3.11+**, [uv](https://docs.astral.sh/uv/) and **Docker**.+**1. Install it and watch the demo.** The demo needs no API keys. It runs the real agent,+sandbox and checks with a scripted model, and renders a launch card in three sizes plus a+4-second teaser into `workspace/out/demo/`.```bashgit clone https://github.com/langchain-ai/design-agent.git@@ -183,108 +82,221 @@ uv run design-agent sandbox builduv run design-agent demo```-The demo needs no keys. A scripted run renders a launch card in three sizes and a 4-second teaser-for the sample brand, through the real agent, sandbox and gates, into `workspace/out/demo/`.--Then add a model and ask for something. Put `ANTHROPIC_API_KEY=…` in `.env` yourself, so the key-stays out of your shell history:+**2. Connect a model.** Put `ANTHROPIC_API_KEY=…` in `.env` yourself, so the key stays out of your+shell history. Then:```bashuv run design-agent config set DESIGN_AGENT_MODEL=anthropic:claude-opus-5-5uv run design-agent test model+```++**3. Ask for something.**++```bashuv run design-agent ask "A LinkedIn card announcing our design review service, 1200x627."```-Or let your coding agent set it up:+Your files land in `workspace/out/<thread>/`, and the last line tells you roughly what the run+cost. Add `--thread <id>` to keep a conversation going.++**Rather have your coding agent set it up?** Open Claude Code, Codex or Cursor in the repo and+paste this:> Read AGENTS.md and .agents/skills/design-agent-onboarding/SKILL.md. Help me connect a model, add> my brand, connect Paper, and choose a deployment path.-## Your brand, media and Paper+## Make it yours-| Add | How | Guide |+Your brand lives in `workspace/brand/`. It holds your colours and the role each one plays, fonts,+logos, voice and graphic language, and it's the only place the agent learns your brand from. The+[brand-onboarding skill](.agents/skills/brand-onboarding/SKILL.md) builds it from your brand guide,+your website, a folder of files or a Paper file. Then `brand validate` and `fonts verify` check it.++| Want | Add | More || --- | --- | --- |-| Your brand | `workspace/brand/` from your guide, site, files or a Paper file; the sample brand until then | [brand-onboarding skill](.agents/skills/brand-onboarding/SKILL.md) |-| Image generation | `OPENAI_API_KEY` | [Models](docs/models.md) |+| Your brand | `workspace/brand/` (until then, the sample brand) | [brand-onboarding skill](.agents/skills/brand-onboarding/SKILL.md) |+| Generated images | `OPENAI_API_KEY` | [Models](docs/models.md) || Voice, music, sound effects | `ELEVENLABS_API_KEY` | [sound skill](skills/sound/SKILL.md) |-| Paper canvas | `PAPER_MODE=local` with Paper Desktop, or `bridged` for a deployed agent | [Paper](docs/partners/paper.md) |-| Web search and other MCP servers | `config/mcp.toml` | [Customization](docs/customization.md) |+| The Paper canvas | `PAPER_MODE=local` with Paper Desktop, or `bridged` for a deployed agent | [Paper](docs/partners/paper.md) |+| Web search, your docs, other tools | MCP servers in `config/mcp.toml` | [Customization](docs/customization.md) |-Licensed fonts go in `private-fonts/` (Git-ignored); open fonts ship in `fonts/`.+Licensed fonts go in `private-fonts/`. Git ignores that folder, so they never get committed.-## Deployment+## Deploy it-Both paths run the same agent core.+Same agent, two ways to run it.-| | Managed Deep Agents · recommended | Self-hosted |+| | Managed Deep Agents (recommended) | Self-hosted || --- | --- | --- |-| Start | `uv run design-agent sandbox publish`, then `uv run design-agent deploy` | `docker compose -f deploy/self-host/compose.yaml up -d --build` |-| Hosting | managed by LangSmith | you operate the API, Postgres and Docker host |-| Sandboxes | managed, cloned per thread from your published snapshot | containers on your Docker host, or LangSmith sandboxes |-| Slack | managed app with workspace authorization | your app, Socket Mode or signed HTTP |+| Get going | `uv run design-agent sandbox publish`, then `uv run design-agent deploy` | `docker compose -f deploy/self-host/compose.yaml up -d --build` |+| Who runs it | LangSmith | you: the API, Postgres and a Docker host |+| Slack | a managed app you authorize | your own Slack app || Guide | [Deploy with MDA](docs/deploy-mda.md) | [Self-hosting](docs/self-hosting.md) |-**Managed Deep Agents** needs a LangSmith organization with MDA access and an API key with-deployment permission. `deploy` stages only what the agent needs, shows which secrets ship, and asks-before deploying; `--name` deploys a separate copy for testing.-[Hosting is paid](https://www.langchain.com/pricing).+`deploy` shows you exactly which settings it will ship and asks before it deploys anything. Want+to try a change without touching your team's agent? Run `deploy --name design-agent-test` and you+get a separate copy. Managed Deep Agents needs a LangSmith organization with MDA access, and+[hosting is paid](https://www.langchain.com/pricing). Model, image and audio usage is billed by your+providers either way.-**Self-hosting** needs Docker. Compose starts Postgres, the API and the sandbox image; Slack is an-optional profile. You operate the infrastructure.+## How it works++Under the hood there are three parts we built for design work, sitting on top of parts we didn't+have to build:++- **The lead and its helpers.** The lead reads your brief, picks the concept and writes a plan.+ It does small jobs itself. Big ones it hands to helpers that work in parallel: three directions+ to compare, or one act each of a film.+- **The context.** Skills, a design wiki, your brand pack and memory. It loads only what the job+ in front of it needs.+- **The designer's computer.** A sandbox with real tools: Chrome, FFmpeg, HyperFrames, D2 and+ your fonts. Everything is built and checked there.+- **Off the shelf:** the [Deep Agents](https://github.com/langchain-ai/deepagents) harness,+ [Managed Deep Agents](https://docs.langchain.com/langsmith/python/managed-deep-agents-overview)+ (or your own infrastructure) and a frontier model.++The split that matters: **the model makes the creative calls, and code does everything that has+to be exactly right.** That means rendering, measuring, checking, delivering files and holding+your keys.++<details>+<summary><b>The architecture, in more detail</b></summary>++One assembly (`src/design_agent/assembly.py`) builds the model, tools, middleware and helpers.+Runtime profiles plug it into Managed Deep Agents, the CLI, the self-hosted API, Slack and+schedules.++**The lead and its helpers**++- Each thread gets one lead. It writes a plan in the thread's project folder and decides what to+ delegate.+- Helpers run in the background on the same sandbox and files. Each one has a heartbeat, a+ durable transcript, and a lead that can steer it.+- Nobody sits and waits. Render watchers and check-ins wake the lead when work lands, fails or+ stalls: one-time schedules on Managed Deep Agents, an in-process waker everywhere else.+- Before anything ships, the lead and helpers look at their own renders with a vision `look`+ tool.++**The context**++- The always-on prompt is small, about 4k tokens. Workflows live in [skills](skills/)+ (`skills/<name>/SKILL.md`) and load on demand. On Managed Deep Agents they mount read-only from+ Context Hub.+- Brand facts come only from the brand pack. It's staged into the sandbox with its fonts and logos.+- Context stays bounded:+ - Long tool results go to files, with a short summary in the conversation.+ - Provider tool search loads MCP tools only when they're needed.+ - Helpers can clear old tool results server-side.+ - Prompt caching keeps the lead's prefix warm for an hour.++**The designer's computer**++- One recipe, [`sandbox/setup.sh`](sandbox/), pins Chrome Headless Shell, FFmpeg, HyperFrames,+ sharp, ECharts, D2 and open fonts by version and sha256. It ends with a probe that renders a+ page, a film and a diagram. The same recipe builds:+ - the local and self-hosted Docker image;+ - the LangSmith snapshot that Managed Deep Agents clones for each thread;+ - the CI image.+- Every thread gets its own sandbox, and its files persist in `/workspace/projects/` between turns.+ - Locally, it's a container with no network, dropped capabilities, an unprivileged user, and+ limits capped to your machine.+ - On Managed Deep Agents, it's a LangSmith sandbox (16 GiB disk, 8 vCPU) with default-deny+ egress.+- Before a run, a manifest with a sha256 checklist stages skills, the checks, your brand pack and+ its fonts into `/workspace`, read-only.++**The checks** ([`workspace/tools/`](workspace/tools/README.md)) run before anything is delivered:++| Check | What it catches |+| --- | --- |+| `brand-lint` | off-brand colours, fonts, radii and copy |+| `video-gate`, `hyperframes check` | non-deterministic animation, layout, motion and contrast problems |+| `artifact-contract` | a wrong size or duration, a missing file, a blank first frame |+| font check | a brand font that silently fell back, web fonts included |+| chart and graph contracts | truncated baselines, mixed units, shares that don't add up to 100%, a diagram that changed shape |++**Keys stay outside.** The sandbox never holds a credential. Host tools run next to the agent,+keep the keys, and drop their results in as files:++- image generation;+- voice, music and sound effects;+- transcription;+- an asset fetcher with SSRF protection;+- Paper, scoped by a policy;+- your MCP servers;+- delivery: to Slack, to `workspace/out/` locally, or attached to the LangSmith trace for API+ runs.++**Where each piece runs**++| | Managed Deep Agents | Self-hosted |+| --- | --- | --- |+| Agent | `agent.py` (`define_deep_agent`), `identity.py`, `tools/mcp.py` | the same assembly in a FastAPI app |+| Threads and memory | LangSmith | Postgres checkpointer and store |+| Sandboxes | LangSmith sandboxes from your snapshot (`sandbox/__init__.py`) | containers on your Docker host |+| Slack | managed app (`channels/slack.py`) | your Bolt app, Socket Mode or signed HTTP |+| Wake-ups and housekeeping | runtime schedules; `schedules/housekeeping.py` every 15 minutes | in-process waker and maintenance loop |+| Skills | synced to Context Hub; binary files ship as deployment files | read from the checkout |++**Models.** Claude Opus 5.5 is the default, and any tool-calling model with vision works. A+registry keyed by exact model id sets effort, tool search and output limits per role: lead,+helper, look, tool selector and learning. The optional [LangSmith Gateway](docs/models.md) uses+each provider's native route, so caching and context editing keep working. Every call is metered+per job, helper and role, and LangSmith traces every run.++Go deeper: [Architecture](docs/architecture/README.md) ·+[Orchestration](docs/architecture/orchestration.md) ·+[Context and failures](docs/architecture/context-and-failures.md).-Model, image and audio charges apply to either path, and `ask` prints each job's estimated spend.+</details>## FAQ-**Can it use my brand instead of LangChain's?** Yes. The brand pack in `workspace/brand/` is the-only source of brand facts. Build it from your guide, website or a Paper file with the-[brand-onboarding skill](.agents/skills/brand-onboarding/SKILL.md), and `brand validate` and-`fonts verify` check it.+**Can it use my brand instead of LangChain's?** Yes, that's the whole point of the brand pack. Add+yours and every output follows it.-**Which models does it support?** Claude Opus 5.5 by default; OpenAI and Google models through-the same registry; any OpenAI-compatible endpoint; and the LangSmith Gateway. See-[Models](docs/models.md).+**Which models does it work with?** Claude Opus 5.5 by default. OpenAI and Google models work+through the same registry, and so do any OpenAI-compatible endpoint and the LangSmith Gateway.+See [Models](docs/models.md).-**Do I need Slack?** No. Use the terminal (`design-agent ask`), the self-hosted HTTP API, or the-LangGraph API of a Managed Deep Agents deployment.+**Do I need Slack?** No. You can use the terminal (`design-agent ask`), the self-hosted HTTP API,+or a Managed Deep Agents deployment's API.-**Can the agent reach the internet or my keys?** The sandbox has no network by default and never-holds a credential. Host tools fetch, generate and deliver on its behalf.+**Can the agent reach the internet or my keys?** No. The sandbox has no network by default and+never holds a credential. Host tools do the fetching, generating and delivering for it.-**How do I know it is any good?** Gates check every file before delivery. The-[eval set](docs/evals.md) grades outputs, copy and behaviours without a model judging taste.+**How do I know the output is any good?** Every file goes through the checks before you get it,+and the [eval set](docs/evals.md) grades outputs, copy and behaviour without a model judging taste.## Built with<p>- <a href="https://github.com/langchain-ai/deepagents"><picture><source media="(prefers-color-scheme: dark)" srcset="docs/assets/readme/logos/deep-agents-dark.svg"><img src="docs/assets/readme/logos/deep-agents-light.svg" alt="Deep Agents" height="28"></picture></a>+ <a href="https://github.com/langchain-ai/deepagents"><picture><source media="(prefers-color-scheme: dark)" srcset="https://github.com/langchain-ai/design-agent/raw/main/docs/assets/readme/logos/deep-agents-dark.svg"><img src="https://github.com/langchain-ai/design-agent/raw/main/docs/assets/readme/logos/deep-agents-light.svg" alt="Deep Agents" height="28"></picture></a> - <a href="https://docs.langchain.com/langsmith/python/managed-deep-agents-overview"><picture><source media="(prefers-color-scheme: dark)" srcset="docs/assets/readme/logos/langsmith-deployment-dark.svg"><img src="docs/assets/readme/logos/langsmith-deployment-light.svg" alt="LangSmith Deployment" height="28"></picture></a>+ <a href="https://docs.langchain.com/langsmith/python/managed-deep-agents-overview"><picture><source media="(prefers-color-scheme: dark)" srcset="https://github.com/langchain-ai/design-agent/raw/main/docs/assets/readme/logos/langsmith-deployment-dark.svg"><img src="https://github.com/langchain-ai/design-agent/raw/main/docs/assets/readme/logos/langsmith-deployment-light.svg" alt="LangSmith Deployment" height="28"></picture></a> <a href="https://hyperframes.heygen.com"><picture><source media="(prefers-color-scheme: dark)" srcset="https://raw.githubusercontent.com/heygen-com/hyperframes/main/docs/logo/dark.svg"><img src="https://raw.githubusercontent.com/heygen-com/hyperframes/main/docs/logo/light.svg" alt="HyperFrames" height="28"></picture></a> <a href="https://paper.design"><img src="https://avatars.githubusercontent.com/u/178361271?s=56" alt="Paper" height="28" align="absmiddle"> <b>Paper</b></a></p>-[Deep Agents](https://github.com/langchain-ai/deepagents) ·-[Managed Deep Agents](https://docs.langchain.com/langsmith/python/managed-deep-agents-overview) ·-[LangSmith](https://smith.langchain.com) · [HyperFrames](https://github.com/heygen-com/hyperframes) ·-[Paper](https://paper.design) · [Chrome Headless Shell](https://developer.chrome.com/blog/chrome-headless-shell) ·+Also: [LangSmith](https://smith.langchain.com) ·+[Chrome Headless Shell](https://developer.chrome.com/blog/chrome-headless-shell) ·[FFmpeg](https://ffmpeg.org) · [D2](https://d2lang.com) · [Apache ECharts](https://echarts.apache.org) ·-[sharp](https://sharp.pixelplumbing.com). Logos belong to their owners; HyperFrames and Paper are+[sharp](https://sharp.pixelplumbing.com). Logos belong to their owners. HyperFrames and Paper arepartner projects we build on.## Build on it-| Change or explore | Start here |+| Want to change or explore | Start here || --- | --- || Instructions, skills and taste | [instructions.md](instructions.md) · [skills/](skills/) · [Customization](docs/customization.md) || Models and cost | [Models](docs/models.md) · [Context and failures](docs/architecture/context-and-failures.md) || Background work, wake-ups, pipelines | [Orchestration](docs/architecture/orchestration.md) |-| The sandbox computer and gates | [sandbox/](sandbox/) · [workspace/tools/](workspace/tools/README.md) |-| Evals and the gallery | [Evals](docs/evals.md) |-| Operations and troubleshooting | [Operations](OPERATIONS.md) |-| Development and tests | [Contributing](CONTRIBUTING.md) · [Agent instructions](AGENTS.md) · [Coding-agent skills](.agents/skills/) |+| The sandbox and the checks | [sandbox/](sandbox/) · [workspace/tools/](workspace/tools/README.md) |+| Evals | [Evals](docs/evals.md) |+| Running it day to day | [Operations](OPERATIONS.md) |+| Contributing | [Contributing](CONTRIBUTING.md) · [Agent instructions](AGENTS.md) · [Coding-agent skills](.agents/skills/) |---