# graph

## Docs

- [Execution model](https://graph.mintlify.app/architecture/execution-model.md): Exactly what runs, and what it costs, per invocation
- [Storage](https://graph.mintlify.app/architecture/storage.md): Plain files by default, abstracted for other backends
- [Changelog](https://graph.mintlify.app/changelog.md)
- [Automation recipes](https://graph.mintlify.app/cookbook/automation.md): Plans as commands: structured output, pipes, and exit codes
- [A custom bot identity](https://graph.mintlify.app/cookbook/bot-identity.md): Give a commenting plan its own name and avatar via a GitHub App
- [CI checks](https://graph.mintlify.app/cookbook/ci-checks.md): Adapt and improve your CI pipelines with ease
- [Reporting](https://graph.mintlify.app/cookbook/reporting.md): Status reports and scheduled digests, annotated
- [Core concepts](https://graph.mintlify.app/getting-started/concepts.md): Plans, the tool catalog, the shape cache, the workbench, and threads
- [Installation](https://graph.mintlify.app/getting-started/installation.md): Release binaries and the container image
- [Introduction](https://graph.mintlify.app/getting-started/introduction.md): A plan engine for repeatable, AI-augmented workflows
- [Quickstart](https://graph.mintlify.app/getting-started/quickstart.md): Install, scaffold a project, run your first plan
- [Models & providers](https://graph.mintlify.app/models/models-and-providers.md): Providers, model roles, named models, and where each inference call goes
- [Agent steps](https://graph.mintlify.app/plans/agent-step.md): A bounded tool-calling loop inside a plan step, with guaranteed structured output
- [Ask steps](https://graph.mintlify.app/plans/ask-step.md): Put a question to the person running the plan, and declare what happens when there isn't one
- [Authoring plans](https://graph.mintlify.app/plans/authoring.md): From blank file to running plan
- [Branching](https://graph.mintlify.app/plans/branching.md): Fork a plan into one of two branches with a decide step
- [Errors & replanning](https://graph.mintlify.app/plans/errors-and-replanning.md): Why your plans never replan, and what plan_and_execute does differently
- [Exit gates](https://graph.mintlify.app/plans/exit-gates.md): End a plan early with a success or error state
- [Finish modes](https://graph.mintlify.app/plans/finish-modes.md): Solver report, structured output, or silence
- [Plan inputs](https://graph.mintlify.app/plans/inputs.md): Schemas, JSON documents, and layered overrides
- [Iteration](https://graph.mintlify.app/plans/iteration.md): Run a body of steps once per item of a list with map and reduce
- [What is a plan](https://graph.mintlify.app/plans/overview.md): Repeatable workflows as reviewable YAML files
- [Selection](https://graph.mintlify.app/plans/selection.md): Partition a list with a per-item gate using filter
- [Template language](https://graph.mintlify.app/plans/template-language.md): The {{Ex.path}} dialect: typed dataflow between steps
- [The planner](https://graph.mintlify.app/plans/the-planner.md): plan_and_execute: LLM-authored plans for novel tasks
- [CLI reference](https://graph.mintlify.app/reference/cli.md): Every command and flag
- [Configuration](https://graph.mintlify.app/reference/configuration.md): config.toml, section by section
- [Plan file schema](https://graph.mintlify.app/reference/plan-schema.md): Every field of a plan document, in one place
- [Scripting contract](https://graph.mintlify.app/reference/scripting-contract.md): Streams, exit codes, JSON envelopes, events, and environment variables
- [Built-ins](https://graph.mintlify.app/tools/builtins.md): Tool packs that ship with Graph
- [graph as an MCP server](https://graph.mintlify.app/tools/mcp-server.md): Serving your plans and the plan authoring commands to another agent
- [MCP servers](https://graph.mintlify.app/tools/mcp-servers.md): Connecting stdio and HTTP MCP servers
- [The tool catalog](https://graph.mintlify.app/tools/overview.md): One namespace for MCP, user, plan, and built-in tools
- [The shape cache](https://graph.mintlify.app/tools/shape-cache.md): How graph learns what tools return
- [User-defined tools](https://graph.mintlify.app/tools/user-defined.md): Wrap commands, queries, and prompts as first-class tools
- [Chat & ask](https://graph.mintlify.app/using/chat-and-ask.md): The agent loop: one-shot questions and the REPL
- [Threads](https://graph.mintlify.app/using/threads.md): Persistent conversations across invocations
- [How the agent picks tools](https://graph.mintlify.app/using/tool-selection.md): Routing between direct calls, your plans, and on-the-fly planning
- [Plan workbench](https://graph.mintlify.app/workbench/plan-workbench.md): A dual-pane TUI for drafting, inspecting, and test-running plans