[Agents] Restructure docs#31098
Open
thomasgauvin wants to merge 6 commits into
Open
Conversation
thomasgauvin
requested review from
a team,
GregBrimble,
MattieTK,
WalshyDev,
elithrar,
irvinebroque,
kodster28,
korinne,
mikenomitch,
rita3ko,
threepointone and
whoiskatrin
as code owners
github-actions
Bot
added
product:workers
Review⏸️ Automatic reviews for this PR are paused. This PR has already received 2 automatic reviews. To run another review, a codeowner can comment
⏸️ Automatic reviews for this PR are paused. This PR has already received 2 automatic reviews. To run another review, a codeowner can comment
💡 1 suggestion found in commit Suggestions (1)
CommandsOnly codeowners can run commands. Post a comment with the command to trigger it.
|
github-actions
Bot
assigned
kodster28,
elithrar,
irvinebroque,
rita3ko,
threepointone,
whoiskatrin,
GregBrimble and
korinne
Contributor
|
This PR requires additional review attention because it affects the following areas: RedirectsThis PR changes current filenames or deletes current files. Make sure you have redirects set up to cover the following paths:
|
Contributor
|
This pull request requires reviews from CODEOWNERS as it changes files that match the following patterns:
|
Contributor
|
CI run failed: build logs |
Contributor
Broken LinksFound 49 broken link(s) across 29 file(s).
|
Contributor
- Rename concept page title to Conversation state and memory (order 4) - Rename MCP top-level title to Model Context Protocol (MCP) - Update RAG title and diagram label to Retrieval Augmented Generation (RAG) - Add redirect for /agents/concepts/memory/
Contributor
irvinebroque
left a comment
irvinebroque
left a comment
There was a problem hiding this comment.
- Diagram is killer 🙌
- A lot of the examples don't use Think, but also don't use another harness. If I'm coming in here and I read the diagram and "get it" that the Agents SDK is the runtime for the harness, and then I read an example, I have questions like, wait - what harness is being used here?
🚀 🚀 🚀 non-blocking, we can keep making better and better (though I do think really showing many harnesses will help orient people, and will help people understand what Think is by having things to reference against, etc.)
| Most AI applications today are stateless — they process a request, return a response, and forget everything. Real agents need more. They need to remember conversations, act on schedules, call tools, coordinate with other agents, and stay connected to users in real-time. The Agents SDK gives you all of this as a TypeScript class. | ||
| Build and host Agents on Cloudflare, connect chat, voice, email, Slack, and webhooks to a durable agent runtime with Browser, Sandbox, AI Search, MCP, Payments and other MCP tools. | ||
|
|
||
| When you host agents on Cloudflare, each agent has a durable identity, local SQL storage, real-time connections, scheduled work, and recoverable execution. |
Contributor
There was a problem hiding this comment.
Suggested change
| When you host agents on Cloudflare, each agent has a durable identity, local SQL storage, real-time connections, scheduled work, and recoverable execution. | |
| When you host agents on Cloudflare, each agent session has a durable identity, local SQL storage, real-time connections, scheduled work, and recoverable execution. |
People use varying language around "agent" vs. "agent session" but it seems like the point we would want to convey is that each "session" or "run" of the agent has its own distinct storage, etc.? Because that is so much more powerful than just "each agent is its own DO" and someone coming away maybe misunderstanding it?
| </div> | ||
| <div class="section-row"> | ||
| <a href="/agents/harnesses/think/">Project Think</a> | ||
| <a href="/agents/runtime/agents-api/">Build-your-own agent</a> |
Contributor
There was a problem hiding this comment.
Seems like having docs for other harnesses will help people truly get it that Project Think is a harness?
Like if I see that there is:
- Project Think
- Pi
- Codex
- Flue
All together then no matter what my frame of reference is for what I am familiar with or not (not everyone will know all of these) then I "get it" even if I am still somewhat shaky on what a harness is, etc.
| @@ -0,0 +1,434 @@ | |||
| --- | |||
Contributor
There was a problem hiding this comment.
love that this diagram is interactive, clever
| Agents on Cloudflare are composed from four parts: | ||
|
|
||
| - **Communication channels** define how users and systems reach your agent, such as [chat](/agents/communication-channels/chat/), [voice](/agents/communication-channels/voice/), [email](/agents/communication-channels/email/), [Slack](/agents/communication-channels/slack/), [webhooks](/agents/communication-channels/webhooks/), and other event sources. | ||
| - **The agent harness** defines the loop: how the agent calls models, selects tools, handles tool results, streams responses, and decides whether to continue. Use [Project Think](/agents/harnesses/think/) for an opinionated harness, or build your own loop directly on the [Agents SDK runtime](/agents/runtime/agents-api/). |
Contributor
There was a problem hiding this comment.
Even if minimal, is there a path to show pi and flue as other harnesses that is possible to use?
Both so that people "get it" on what Think is and so that people understand that building agents on Cloudflare isn't "use Think or build the whole thing yourself" - I feel like i have options if presented to me, and therefore more comfortable betting on the Cloudflare-native way in Think
| <AgentsPlatformDiagram /> | ||
|
|
||
| The mental model is simple: define a TypeScript class, give each real-world thing a stable name, and route requests or WebSocket connections to that named instance. The instance wakes when something happens, reads its durable state, does work, and hibernates when idle. | ||
| Agents on Cloudflare are composed from four parts: |
Contributor
There was a problem hiding this comment.
This part is so strong and clear that might even lead with it
| <AgentsPlatformDiagram /> | ||
|
|
||
| The mental model is simple: define a TypeScript class, give each real-world thing a stable name, and route requests or WebSocket connections to that named instance. The instance wakes when something happens, reads its durable state, does work, and hibernates when idle. | ||
| Agents on Cloudflare are composed from four parts: |
Contributor
There was a problem hiding this comment.
I do think we are maybe missing inference from this? i.e. — that you can use any model, that there is AI Gateway that supports hosted Workers AI models and hosted models, etc.
| The starter includes streaming AI chat, server-side and client-side tools, human-in-the-loop approval, and task scheduling — a foundation you can build on or tear apart. You can also swap in [OpenAI, Anthropic, Google Gemini, or any other provider](/agents/runtime/operations/using-ai-models/). | ||
|
|
||
| <TypeScriptExample> | ||
| ### Example agents |
Contributor
There was a problem hiding this comment.
YES - love putting this front and center here with clear use cases
|
|
||
| Check the [GitHub page](https://github.com/cloudflare/mcp-server-cloudflare) to learn how to use Cloudflare's remote MCP servers with different MCP clients. | ||
|
|
||
| ## Cloudflare Community MCP server |
Contributor
There was a problem hiding this comment.
I'm not quite sure I follow why this is here?
We should probably update these docs to be clearer that the Code Mode MCP server is the MCP server and de-emphasize the rest?
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
11 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.