# Redgold Documentation > Documentation for Redgold, a platform for building typed pipeline applications with managed data, agents, models, and deployment Machine-readable index of the public Redgold documentation, following the llms.txt convention (https://llmstxt.org). https://docs.redgold.ai/llms-full.txt carries the full markdown of every page listed here. ## Intro - [Quickstart](https://docs.redgold.ai/intro/quickstart): Start one continuing project in Redgold, finish its first useful task, and carry it into users, payments, testing, publishing, and growth. - [Pipeline Applications](https://docs.redgold.ai/intro/pipeline-app-walkthrough): See the customer-controlled artifacts and managed runtime behind a Redgold pipeline application. - [How Redgold works](https://docs.redgold.ai/intro/overview): Build and maintain applications with persistent AI coding agents, a compatible model API, and managed data and deployment. - [Positioning](https://docs.redgold.ai/intro/comparison/positioning): Where Redgold fits across application deployment, data processing, model access, and agent operations. - [vs Lovable](https://docs.redgold.ai/intro/comparison/vs-lovable): Redgold against Lovable and similar AI app builders across generation, backend, data, and infrastructure ownership. - [vs Cloudflare OS](https://docs.redgold.ai/intro/comparison/vs-cloudflare-os): Redgold against Cloudflare OS across agent workspaces, personal modifiable apps, sandboxing, and infrastructure ownership. - [vs a backend-as-a-service](https://docs.redgold.ai/intro/comparison/vs-baas): Redgold against Supabase, Firebase, and similar hosted app backends. - [vs serverless compute](https://docs.redgold.ai/intro/comparison/vs-serverless-compute): Redgold against Modal and similar serverless-compute platforms. - [vs a do-it-yourself data stack](https://docs.redgold.ai/intro/comparison/vs-diy-data-stack): Redgold against a self-assembled Kubernetes, Postgres, DuckDB, and glue stack. - [vs Rama](https://docs.redgold.ai/intro/comparison/vs-rama): Redgold against Rama across programming model, storage, execution, deployment, and agent integration. - [vs Apache Spark](https://docs.redgold.ai/intro/comparison/vs-spark): Redgold against Apache Spark across data processing, storage, serving, deployment, and agent integration. - [vs Apache DataFusion](https://docs.redgold.ai/intro/comparison/vs-datafusion): Redgold against Apache DataFusion across query execution, storage, application serving, and operations. - [vs Heroku](https://docs.redgold.ai/intro/comparison/vs-heroku): Redgold against Heroku across deployment, infrastructure ownership, data, and agent operations. - [vs AI model providers](https://docs.redgold.ai/intro/comparison/vs-ai-providers): Redgold against direct OpenAI, Anthropic, Google, and similar model APIs. - [Common tasks](https://docs.redgold.ai/intro/starting-points): Find agent tools, prompt and MCP setup, remote builds, or the next step in building and operating a project. - [FAQ](https://docs.redgold.ai/intro/faq): Pipeline apps, agent guardrails, model compatibility, opt-in scope, and API documentation boundaries. ## Cookbook - [Examples](https://docs.redgold.ai/cookbook/overview): Task-oriented recipes — one page per outcome, each with the commands, the expected output, and where to go when it breaks. - [Run a one-shot agent task in CI](https://docs.redgold.ai/cookbook/one-shot-ci): Drive a single agent task to completion with `redgold --one-shot` and gate on its exit code. - [Retry model API calls under rate limits](https://docs.redgold.ai/cookbook/retry-rate-limits): Handle 429, 402, and 502 the way the model API expects — honor Retry-After, back off with jitter, and correct what you should not retry. - [Connect GitHub review automation](https://docs.redgold.ai/cookbook/connect-github-app): Prepare a repository for Redgold GitHub App review and fixer automation during managed onboarding. - [Delegate a task to another agent](https://docs.redgold.ai/cookbook/delegate-agent-task): Send bounded work to an assisting agent through customer REST control or an authorized managed agent-cli environment. - [Validate a pipeline application change](https://docs.redgold.ai/cookbook/validate-pipeline-app): Review the schema, manifest, transform, and public contract of a narrow application before publication. - [Verify a managed application deployment](https://docs.redgold.ai/cookbook/verify-managed-deploy): Follow a reviewed application change from repository checks to a managed deployment without relying on cluster access. - [Build, launch, and grow one project](https://docs.redgold.ai/cookbook/launch-and-grow): Carry an application from its first working feature through users, payments, tests, web delivery, mobile-store preparation, marketing, and ongoing improvements. - [Set up managed reviews and project memory](https://docs.redgold.ai/cookbook/managed-review): Plan a paid managed review pilot, configure GitHub review behavior and retained project context, and inspect the first completed result. - [Create an API key and make your first call](https://docs.redgold.ai/cookbook/first-api-call): Mint an sk-rg key, put it in your environment, and confirm it against the authenticated model list. - [Call the model API from Python](https://docs.redgold.ai/cookbook/model-api-python): Drive /v1/chat/completions with the OpenAI SDK pointed at api.redgold.ai. - [Use the Anthropic Messages API](https://docs.redgold.ai/cookbook/anthropic-messages): Point the Anthropic SDK at api.redgold.ai and authenticate with a bearer token instead of x-api-key. - [Check your model-API credit balance](https://docs.redgold.ai/cookbook/check-credits): Read /v1/credits so a long run does not fail partway through on a 402. - [Connect an MCP client and query your data](https://docs.redgold.ai/cookbook/mcp-query-data): Discover an accessible dataset and run a bounded query with durable results. - [Run a reviewed pipeline from an MCP client](https://docs.redgold.ai/cookbook/mcp-drive-agent): Prepare, run and inspect a bounded transformation with durable outputs. - [Install the terminal client and agent tools](https://docs.redgold.ai/cookbook/install-client): Install the Redgold terminal client, RAC and its CI companion, check service access and keep the tools updated. - [Point the terminal client at another API host](https://docs.redgold.ai/cookbook/point-client-at-host): Connect the Redgold terminal client to a self-hosted instance with --api or REDGOLD_API. ## Fleet - [Running agent work](https://docs.redgold.ai/fleet/overview): Organize and supervise the agents that build and maintain your managed applications. - [Workstreams and issues](https://docs.redgold.ai/fleet/workstreams-and-issues): Turn an application objective into reviewable work with dependencies and delivery evidence. - [Agent lifecycle](https://docs.redgold.ai/fleet/agent-lifecycle): Understand assignment, startup, waiting, recovery, and completion in managed agent work. - [Supervision and recovery](https://docs.redgold.ai/fleet/supervision): Inspect fleet progress, resolve blockers, and recover managed application work. - [An application workstream from request to delivery](https://docs.redgold.ai/fleet/application-workstream): A worked support-app change spanning issue tracking, agents, a blocked decision, and managed deployment. ## Users - [Connect your existing agent](https://docs.redgold.ai/users/install/agent-tools): Add search, browsing, memory, managed compute or remote builds to your existing agent with a setup prompt, MCP or the RAC CLI. ## Infrastructure - [Architecture](https://docs.redgold.ai/infrastructure/overview): How Redgold builds, runs, and evolves fully managed pipeline applications. - [A request through the platform](https://docs.redgold.ai/infrastructure/request-path): Follow a managed application request through identity, execution, storage, and response. - [Agent work and durable state](https://docs.redgold.ai/infrastructure/agent-work): How application objectives, agent execution, and managed delivery stay connected across sessions. - [Data and pipelines](https://docs.redgold.ai/infrastructure/data): How agent activity and application records move through shared datasets and pipeline execution. - [AI and agents](https://docs.redgold.ai/infrastructure/ai): Coding-agent workspaces, standalone agent tools, compatible model APIs, persistence and replay boundaries. - [Managed deployment](https://docs.redgold.ai/infrastructure/deployment): How a reviewed application change becomes a verified live revision on Redgold. ## Reference - [Authentication](https://docs.redgold.ai/reference/authentication): Account sessions for interactive clients and sk-rg API keys for SDKs, scripts, and CI. - [Security disclosure policy](https://docs.redgold.ai/reference/security-disclosure): How to report a security vulnerability in Redgold, what is in scope, our safe-harbor commitment, and the response you can expect. - [RAC command guide](https://docs.redgold.ai/reference/rac): Current RAC workflows for search, browsers, memory, remote builds, data, integrations and managed agent work. - [Model API](https://docs.redgold.ai/reference/llm-api): OpenAI- and Anthropic-compatible endpoints serving Redgold models at api.redgold.ai. - [MCP server](https://docs.redgold.ai/reference/mcp): Connect an external MCP client to the hosted API, data and reviewed pipelines. - [Data platform preview](https://docs.redgold.ai/reference/data-platform-preview): The supported Developer Preview contract for owned datasets, typed Rust records, queries, and pipeline-backed applications. - [Terminal client](https://docs.redgold.ai/reference/cli): Install, authenticate and use the Redgold terminal conversation client alongside RAC tools. - [Pipeline manifest preview](https://docs.redgold.ai/reference/pipeline-preview): A public Developer Preview of the Git-authored pipeline manifest used to define routes, transforms, datasets, and execution order. - [Limits and retries](https://docs.redgold.ai/reference/limits): Current model API rate-limit, upstream error, and client retry behavior. - [API contract details](https://docs.redgold.ai/reference/api-contracts): Generated request, response, streaming, error, limit, compatibility, and example details for supported public API routes. - [Errors](https://docs.redgold.ai/reference/errors): Provider-compatible model API errors and practical retry guidance. - [Memory routes](https://docs.redgold.ai/reference/routes/memory): Generated reference for the supported Memory API routes. - [V1 routes](https://docs.redgold.ai/reference/routes/v1): Generated reference for the supported V1 API routes. - [Voice routes](https://docs.redgold.ai/reference/routes/voice): Generated reference for the supported Voice API routes. ## Develop - [Author a batch transform](https://docs.redgold.ai/develop/author-a-batch-transform): Scaffold, check, publish, and pin a trusted Rust transform that processes Arrow IPC batches in the dataflow engine. - [Pipeline apps](https://docs.redgold.ai/develop/pipeline-apps/overview): Portable application contracts, strict TOML compilation, and a shared declarative host built on the pipeline backend. - [Collaborative board walkthrough](https://docs.redgold.ai/develop/pipeline-apps/collaborative-board): A developer-preview portable pipeline app spanning release distribution, rendering, durable replay, and causal entity merge. - [Release and security lifecycle](https://docs.redgold.ai/develop/pipeline-apps/release-lifecycle): How portable pipeline apps are hashed, signed, validated, installed, permissioned, upgraded, shared, forked, and revoked. - [Incident Triage](https://docs.redgold.ai/develop/pipeline-apps/incident-triage): A second portable application using the same signed entities, actions and shared host as the collaborative board. - [App layout and authoring previews](https://docs.redgold.ai/develop/pipeline-apps/presentation): Compile declarative templates and scoped CSS into portable app components, then review real desktop and mobile rendering. - [Metrics Notebook](https://docs.redgold.ai/develop/pipeline-apps/metrics-notebook): A gallery app that saves named metric queries and runs them through signed, parameterized pipelines on the shared host. - [Webhook Relay](https://docs.redgold.ai/develop/pipeline-apps/webhook-relay): A gallery app that records inbound webhook events, matches them to relay rules and tracks each delivery's status on the shared host. - [Tasks, repositories and applications from RAC](https://docs.redgold.ai/develop/agent-task-cli): Use RAC from a laptop, CI job or agent with your platform identity. - [Build a vertical](https://docs.redgold.ai/develop/build-a-vertical): How to build an application vertical on the platform, in order — Rust schema, ingestion, transforms, service surface, UI, then deployment. - [Managed application artifacts](https://docs.redgold.ai/develop/scaffold-a-narrow-app): The schema, pipeline TOML, Rust transform, tests, and API contract reviewed for a managed pipeline application. - [Import your data](https://docs.redgold.ai/develop/import-your-data): Import JSONL or Parquet into an existing declared dataset, resume acknowledged batches and verify owner-scoped rows. ## Applications - [Application maturity catalog](https://docs.redgold.ai/applications/catalog): Generated status and access summary for the public Redgold application documentation. - [Coding agents](https://docs.redgold.ai/applications/ai-agents/overview): Workspaces, sessions, tools, review, and bounded delegation for repository tasks. - [GitHub App repository automation](https://docs.redgold.ai/applications/ai-agents/github-app-automation): Repository-scoped pull-request review and correction for repositories onboarded to the Redgold GitHub App. - [Assist agents with agent-cli](https://docs.redgold.ai/applications/ai-agents/agent-cli-assistance): Managed fleet workflows for agents and operators with assigned compute and the corresponding permissions. - [Control an assigned agent through REST](https://docs.redgold.ai/applications/ai-agents/rest-agent-control): Start, prompt, and stop an agent slot assigned to your account through the authenticated V1 API. - [Tickets](https://docs.redgold.ai/applications/tickets/overview): The issue-board app, walked end to end — one checked-in Rust serde schema encoded with CBOR, a set of pipeline routes, owner-scoped storage, and a GitOps deploy — as the worked example of the platform's narrow-app shape. - [Tickets API walkthrough](https://docs.redgold.ai/applications/tickets/api-walkthrough): Create, read, comment on, claim, release, and close an owner-scoped ticket through the managed preview API. - [Case study: owner-scoped tickets](https://docs.redgold.ai/applications/case-studies/tickets): How the tickets application combines typed records, narrow routes, lifecycle transitions, and agent dispatch. - [Case study: repository review and correction](https://docs.redgold.ai/applications/case-studies/repository-review): How an onboarded GitHub repository turns pull-request events into scoped agent reviews and eligible fixes. - [Case study: repairing crawlability on Redgold’s website](https://docs.redgold.ai/applications/case-studies/public-website-seo): How Redgold repaired soft 404s, added share metadata, and checked public pages while preserving application entry routes. - [Workflow account: Elusive reviews and managed CI](https://docs.redgold.ai/applications/case-studies/elusive-managed-review): How Elusive uses Redgold repository reviews and remote Linux CI, with observed review evidence and clear limits on savings and quality claims. ## Resources - [Glossary](https://docs.redgold.ai/resources/glossary): Public terms used across the Redgold pipeline-application platform and its supporting capabilities. - [Terms of Service](https://docs.redgold.ai/resources/terms): The terms that govern use of the Redgold platform. - [Privacy Policy](https://docs.redgold.ai/resources/privacy): What Redgold stores, how it is used, and your deletion rights. - [Acceptable Use Policy](https://docs.redgold.ai/resources/acceptable-use): What you may and may not do on the Redgold platform. - [FAQ](https://docs.redgold.ai/resources/faq): Common questions about accounts, models, agents, data, billing, privacy, and self-hosting. - [Documentation feedback](https://docs.redgold.ai/resources/docs-feedback): Report a missing, unclear, or outdated documentation page through Redgold's first-party support channel. - [Refund and Cancellation Policy](https://docs.redgold.ai/resources/refund-cancellation): How subscriptions, prepaid credits, cancellations, and refund requests are handled. ## Overview - [Home](https://docs.redgold.ai/): ui: container: pt-12!