From 9fc9a0e4e6bf460697b6f08b4481befe7267e29f Mon Sep 17 00:00:00 2001 From: "warp-agent-staging[bot]" <240773466+warp-agent-staging[bot]@users.noreply.github.com> Date: Fri, 9 Oct 2026 19:06:01 +0000 Subject: [PATCH 1/3] docs: connect orchestration patterns to Warp coordination features --- src/content/docs/platform/orchestration/index.mdx | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/src/content/docs/platform/orchestration/index.mdx b/src/content/docs/platform/orchestration/index.mdx index 65e1a3791..fd1cc5f8b 100644 --- a/src/content/docs/platform/orchestration/index.mdx +++ b/src/content/docs/platform/orchestration/index.mdx @@ -7,7 +7,7 @@ sidebar: import VideoEmbed from '@components/VideoEmbed.astro'; import { VARS } from '@data/vars'; -Multi-agent orchestration lets one agent spawn and coordinate other agents to parallelize work, delegate specialized tasks, or verify another agent's output. The parent/child model works from the Warp app, the [{VARS.WARP_AGENT_CLI}](/agents/cli/oz-cli/), and the [{VARS.WARP_PLATFORM_API}](/factories/api-and-sdk/), and supports local, cloud, and mixed execution. +Warp coordinates multiple agents through a parent agent that spawns child agents to work in parallel, handle specialized tasks, or verify output. The parent/child model works from the Warp app, the [{VARS.WARP_AGENT_CLI}](/agents/cli/oz-cli/), and the [{VARS.WARP_PLATFORM_API}](/factories/api-and-sdk/), and supports local, cloud, and mixed execution. Watch this walkthrough to see how a cloud agent can coordinate a team of agents in the cloud. @@ -83,7 +83,14 @@ Use messages for coordination signals - handoffs, decisions, blocked-on-input re ## Common patterns -The following patterns show common ways to structure parent and child agents, depending on whether you need parallel execution, review, dependency ordering, or loose coordination. +| Pattern | When to use it | Warp feature behind the pattern | +| --- | --- | --- | +| [Supervisor and worker](#supervisor--worker) | Divide a task into work items with one agent owning coordination. | A [parent agent](#the-parentchild-model) assigns work to children and collects results. | +| [Fan-out and fan-in](#fan-out--fan-in) | Run a refactor or migration across independent modules or targets. | Parallel child runs and [lifecycle events](#run-state-transitions) let the parent wait and combine results. | +| [Critic and verifier](#critic--verifier) | Check a proposed solution before accepting it. | [Messaging](#messaging-between-agents) carries review notes and follow-up instructions between writer and critic. | +| [Review swarm](#review-swarm-cloud--cloud) | Review multiple open PRs from a scheduled or webhook-triggered run. | [Cloud children](#where-parent-and-child-agents-can-run) run reviews in parallel; the parent collects their findings. | +| [DAG](#dag) | Subtasks must wait for upstream results, such as build, test, then deploy. | The parent observes child [state transitions](#run-state-transitions) before starting dependent work. | +| [Swarm](#swarm) | Peer agents need loose coordination rather than a strict hierarchy. | Agent [inboxes and messaging](#messaging-between-agents) carry coordination signals; the parent remains the coordinator. | ### Supervisor / worker From efaa61faad467abcf2cfbc7d3c1e170fbe97fe7d Mon Sep 17 00:00:00 2001 From: "warp-agent-staging[bot]" <240773466+warp-agent-staging[bot]@users.noreply.github.com> Date: Fri, 9 Oct 2026 19:36:04 +0000 Subject: [PATCH 2/3] docs: distinguish critic output from coordination messaging --- src/content/docs/platform/orchestration/index.mdx | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/src/content/docs/platform/orchestration/index.mdx b/src/content/docs/platform/orchestration/index.mdx index fd1cc5f8b..c676a14e0 100644 --- a/src/content/docs/platform/orchestration/index.mdx +++ b/src/content/docs/platform/orchestration/index.mdx @@ -7,7 +7,7 @@ sidebar: import VideoEmbed from '@components/VideoEmbed.astro'; import { VARS } from '@data/vars'; -Warp coordinates multiple agents through a parent agent that spawns child agents to work in parallel, handle specialized tasks, or verify output. The parent/child model works from the Warp app, the [{VARS.WARP_AGENT_CLI}](/agents/cli/oz-cli/), and the [{VARS.WARP_PLATFORM_API}](/factories/api-and-sdk/), and supports local, cloud, and mixed execution. +Warp uses multi-agent orchestration to let a parent agent spawn child agents to work in parallel, handle specialized tasks, or verify output. The parent/child model works from the Warp app, the [{VARS.WARP_AGENT_CLI}](/agents/cli/oz-cli/), and the [{VARS.WARP_PLATFORM_API}](/factories/api-and-sdk/), and supports local, cloud, and mixed execution. Watch this walkthrough to see how a cloud agent can coordinate a team of agents in the cloud. @@ -87,22 +87,22 @@ Use messages for coordination signals - handoffs, decisions, blocked-on-input re | --- | --- | --- | | [Supervisor and worker](#supervisor--worker) | Divide a task into work items with one agent owning coordination. | A [parent agent](#the-parentchild-model) assigns work to children and collects results. | | [Fan-out and fan-in](#fan-out--fan-in) | Run a refactor or migration across independent modules or targets. | Parallel child runs and [lifecycle events](#run-state-transitions) let the parent wait and combine results. | -| [Critic and verifier](#critic--verifier) | Check a proposed solution before accepting it. | [Messaging](#messaging-between-agents) carries review notes and follow-up instructions between writer and critic. | +| [Critic and verifier](#critic--verifier) | Check a proposed solution before accepting it. | The critic's output returns notes to the parent; the parent sends follow-up instructions through [messaging](#messaging-between-agents). | | [Review swarm](#review-swarm-cloud--cloud) | Review multiple open PRs from a scheduled or webhook-triggered run. | [Cloud children](#where-parent-and-child-agents-can-run) run reviews in parallel; the parent collects their findings. | | [DAG](#dag) | Subtasks must wait for upstream results, such as build, test, then deploy. | The parent observes child [state transitions](#run-state-transitions) before starting dependent work. | -| [Swarm](#swarm) | Peer agents need loose coordination rather than a strict hierarchy. | Agent [inboxes and messaging](#messaging-between-agents) carry coordination signals; the parent remains the coordinator. | +| [Swarm](#swarm) | Coordinate a flat group of agents without a strict hierarchy. | The parent acts as a coordinator; agents use [messaging](#messaging-between-agents) for coordination signals. | ### Supervisor / worker -A parent supervisor agent breaks the task into a queue of work items, spawns worker children to claim and complete each item, and writes a summary when the queue is empty. Use this when the task is naturally divisible and you want a single agent to own coordination. +A parent supervisor agent breaks the task into a queue of work items, spawns worker children to claim and complete each item, and writes a summary when the queue is empty. ### Fan-out / fan-in -The parent spawns N children in parallel, each with a sharded prompt (one module, one file set, one test target, one model), then waits for all of them to complete and merges their results. Use this for large refactors, repo-wide migrations, or running the same task across multiple targets. +The parent spawns N children in parallel, each with a sharded prompt (one module, one file set, one test target, one model), then waits for all of them to complete and merges their results. ### Critic / verifier -The parent (the "writer") proposes a solution, then spawns a critic child to review it. The critic returns notes; the writer revises; the cycle repeats until the critic approves or a budget is exhausted. Useful when correctness matters more than throughput. +The parent (the "writer") proposes a solution, then spawns a critic child to review it. The critic returns notes; the writer revises; the cycle repeats until the critic approves or a budget is exhausted. ### Review swarm (cloud → cloud) @@ -110,7 +110,7 @@ A [scheduled](/platform/triggers/scheduled-agents/) or webhook-triggered parent ### DAG -The parent encodes a directed acyclic graph of subtasks where some nodes depend on the outputs of others. It spawns ready nodes, waits on their state transitions, and spawns dependents as upstream nodes complete. Use this when the workflow has explicit ordering constraints (build → test → deploy, for example). +The parent encodes a directed acyclic graph of subtasks where some nodes depend on the outputs of others. It spawns ready nodes, waits on their state transitions, and spawns dependents as upstream nodes complete. ### Swarm From 2a361d7fef6b2d2e67b736ef8025ba54f9d34dff Mon Sep 17 00:00:00 2001 From: "warp-agent-staging[bot]" <240773466+warp-agent-staging[bot]@users.noreply.github.com> Date: Wed, 14 Oct 2026 16:14:14 +0000 Subject: [PATCH 3/3] Address agent docs review: direct opening, table header, critic trade-off --- src/content/docs/platform/orchestration/index.mdx | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/src/content/docs/platform/orchestration/index.mdx b/src/content/docs/platform/orchestration/index.mdx index 26945fef1..dd9347364 100644 --- a/src/content/docs/platform/orchestration/index.mdx +++ b/src/content/docs/platform/orchestration/index.mdx @@ -7,7 +7,7 @@ sidebar: import VideoEmbed from '@components/VideoEmbed.astro'; import { VARS } from '@data/vars'; -Warp uses multi-agent orchestration to let a parent agent spawn child agents to work in parallel, handle specialized tasks, or verify output. The parent/child model works from the Warp app, the [{VARS.WARP_AGENT_CLI}](/agents/cli/oz-cli/), and the [{VARS.WARP_PLATFORM_API}](/factories/api-and-sdk/), and supports local, cloud, and mixed execution. +Multi-agent orchestration in Warp lets a parent agent spawn and coordinate child agents to work in parallel, handle specialized tasks, or verify output. The parent/child model works from the Warp app, the [{VARS.WARP_AGENT_CLI}](/agents/cli/oz-cli/), and the [{VARS.WARP_PLATFORM_API}](/factories/api-and-sdk/), and supports local, cloud, and mixed execution. Watch this walkthrough to see how a cloud agent can coordinate a team of agents in the cloud. @@ -83,11 +83,11 @@ Use messages for coordination signals - handoffs, decisions, blocked-on-input re ## Common patterns -| Pattern | When to use it | Warp feature behind the pattern | +| Pattern | When to use it | What Warp provides | | --- | --- | --- | | [Supervisor and worker](#supervisor--worker) | Divide a task into work items with one agent owning coordination. | A [parent agent](#the-parentchild-model) assigns work to children and collects results. | | [Fan-out and fan-in](#fan-out--fan-in) | Run a refactor or migration across independent modules or targets. | Parallel child runs and [lifecycle events](#run-state-transitions) let the parent wait and combine results. | -| [Critic and verifier](#critic--verifier) | Check a proposed solution before accepting it. | The critic's output returns notes to the parent; the parent sends follow-up instructions through [messaging](#messaging-between-agents). | +| [Critic and verifier](#critic--verifier) | Check a proposed solution before accepting it, when correctness matters more than throughput. | The critic child reports its notes in its final output, and the parent sends follow-up instructions through [messaging](#messaging-between-agents). | | [Review swarm](#review-swarm-cloud--cloud) | Review multiple open PRs from a scheduled or webhook-triggered run. | [Cloud children](#where-parent-and-child-agents-can-run) run reviews in parallel; the parent collects their findings. | | [DAG](#dag) | Subtasks must wait for upstream results, such as build, test, then deploy. | The parent observes child [state transitions](#run-state-transitions) before starting dependent work. | | [Swarm](#swarm) | Coordinate a flat group of agents without a strict hierarchy. | The parent acts as a coordinator; agents use [messaging](#messaging-between-agents) for coordination signals. |