# dagraph

## Docs

- [Approve gates and respond to user_input nodes via CLI](https://docs.dagraph.app/cli/approve-respond.md): Unblock paused dagraph runs by approving or rejecting approval_gate nodes, and supply answers to user_input nodes, directly from the CLI.
- [Visualize DAG topology and diff artifacts across runs](https://docs.dagraph.app/cli/graph-diff.md): Render a DAG as Mermaid or Graphviz to understand its structure, then compare per-node artifacts between two runs for prompt iteration or A/B testing.
- [Inspect node statuses and run artifacts in dagraph](https://docs.dagraph.app/cli/inspect.md): View per-node execution status, artifact digests, and full output content for any completed, partial, or paused dagraph run using agentgraph inspect.
- [List dagraph runs and resume incomplete ones quickly](https://docs.dagraph.app/cli/list-resume.md): Audit all dagraph runs in your runs directory with status filtering, and resume paused or interrupted runs without re-running completed nodes.
- [agentgraph CLI reference: all 11 commands explained](https://docs.dagraph.app/cli/overview.md): Full reference for all agentgraph CLI commands: run, validate, inspect, approve, respond, list, resume, graph, diff, serve, and schedule.
- [agentgraph run: execute DAG workflows from the CLI](https://docs.dagraph.app/cli/run.md): Execute a dagraph YAML workflow from the CLI. Control the backend, sandbox, concurrency, rate limits, and deterministic replay with a single command.
- [Trigger dagraph workflows via webhook or cron schedule](https://docs.dagraph.app/cli/serve-schedule.md): Run dagraph workflows automatically on a cron schedule or in response to HTTP POST requests using the serve and schedule commands.
- [Validate a DAG file and preview the execution plan](https://docs.dagraph.app/cli/validate.md): Parse a DAG YAML, verify graph structure, preview the execution wave plan, and catch undefined template variables before spending any tokens.
- [Artifacts, run outputs, and deterministic replay in dagraph](https://docs.dagraph.app/concepts/artifacts.md): How dagraph stores node outputs in a content-addressed store, injects them in downstream prompts, writes output files, and replays runs without LLM calls.
- [LLM backends and per-node model routing in dagraph](https://docs.dagraph.app/concepts/backends.md): Configure dagraph's 7 LLM backends (Claude Code, Anthropic API, OpenAI, Gemini, Bedrock, Ollama, Codex) and pin any node to a provider with a model prefix.
- [dagraph workflow YAML: structure, fields, and syntax](https://docs.dagraph.app/concepts/dag-yaml.md): Learn how to write a dagraph YAML workflow file: top-level fields, input declarations, node lists, output mappings, and Jinja2 template variables.
- [All dagraph node types: reference and YAML examples](https://docs.dagraph.app/concepts/node-types.md): Reference for all 10 dagraph node types: agent, evaluator_loop, loop, map, planner, bash, python_exec, subgraph, approval_gate, and user_input.
- [Control workflow costs with budgets and retry policies](https://docs.dagraph.app/configuration/budget-retry.md): Set hard token and USD caps at the DAG or node level to prevent cost runaway, and configure per-node retry policies to handle transient provider errors.
- [dagraph DAG YAML: complete schema and field reference](https://docs.dagraph.app/configuration/dag-schema.md): Every top-level field in a dagraph YAML file: name, description, inputs, budget, nodes, outputs, and hooks — with types, defaults, and examples.
- [Configure lifecycle hooks for dagraph workflow events](https://docs.dagraph.app/configuration/lifecycle-hooks.md): Fire webhook POSTs or shell commands when a DAG starts, completes, pauses, or fails — and when individual nodes start, complete, or fail.
- [Configure agent memory scopes and run tracing in dagraph](https://docs.dagraph.app/configuration/memory-tracing.md): Persist conversation history across runs with named memory scopes, and inspect token usage and timing with OTel-compatible JSON Lines traces.
- [Iterative AI refinement with dagraph evaluator loops](https://docs.dagraph.app/guides/evaluator-loops.md): Pair a generator with a separate evaluator model that scores each attempt and feeds back critique, looping until the output meets your quality bar.
- [Run bash and Python exec nodes in dagraph workflows](https://docs.dagraph.app/guides/exec-nodes.md): Mix bash and python_exec nodes with LLM agent calls in a single DAG — collect system data, transform outputs, and pass results downstream as artifacts.
- [Add human approval gates to your dagraph workflows](https://docs.dagraph.app/guides/human-in-the-loop.md): Pause a dagraph workflow so a human can inspect artifacts, approve or reject the result, then resume execution — with full CLI commands for each step.
- [Configure multi-provider fallback chains in dagraph](https://docs.dagraph.app/guides/multi-provider-fallback.md): Keep workflows running through provider outages and rate limits by routing each node through an ordered fallback chain of alternative models.
- [Run parallel AI agents across your dagraph workflow](https://docs.dagraph.app/guides/parallel-agents.md): Fan out work to multiple AI agents that run simultaneously, then merge their outputs — cutting wall-clock time and token costs with wave-based execution.
- [Schedule and webhook-trigger your dagraph workflows](https://docs.dagraph.app/guides/scheduling-webhooks.md): Run dagraph workflows on a cron schedule, trigger them via HTTP webhook, or combine both in one long-running server process with optional Bearer auth.
- [Install dagraph: pip, uv, and backend configuration](https://docs.dagraph.app/installation.md): Requirements, pip and uv install options, source install, and step-by-step setup for all 7 LLM backends including claude_code, OpenAI, Gemini, and Bedrock.
- [dagraph: run parallel AI agent workflows from YAML](https://docs.dagraph.app/introduction.md): dagraph is a workflow-as-code CLI that runs YAML-defined DAGs of LLM calls in parallel across 7 providers, with HITL gates and deterministic replay.
- [Get started with dagraph: run your first workflow fast](https://docs.dagraph.app/quickstart.md): Install dagraph, write a two-node workflow, and run your first parallel agent pipeline in under 5 minutes. No API key required for the default backend.