docs: add a guide for connecting your own AI agent over MCP - #8241
Conversation
States capability and permissions once instead of repeating them, drops the prerequisites framing in favour of the agent resolving version issues, and reframes the editor session as agent-guided.
|
Reworked from the review. The structural point drove most of it: capability and permissions were restated in four places, so they are now stated once and everything else refers to that. Prerequisites removed. No "before you start", no version checklist. Versions moved to a closing "Getting the best out of it" section that opens by saying this is not something to set up in advance, because the agent tells you what is in the way and helps resolve it. Self-hosted broker is now "depends on how your platform was installed" rather than a setup step, per FlowFuse/helm#1010. Wording. AI agent throughout, never assistant. FlowFuse applications, and flows inside your Node-RED instances. Editing rights rather than changes. Dropped "deliberately" and "exposure"; the editor session is agent-guided, with the header control for indicating which browser session it works in. Added the point that your own agent means your own model or provider, in both this guide and the Expert index. Softened. No-delete and manual deploy both carry "for now" plus the intent about acting as required in production and in setups where experimentation is permitted. Read access now simply has no ability to change anything. Tool annotations are described as recommended usage and permissions. Two things to note:
|
|
Up for review |
OAuth is stated as the intended route and the token route is kept for the clients that cannot sign in, split into the two shapes that actually exist: a header, and a local command via mcp-remote for Claude Desktop. Also from review: HTTP MCP support is the real requirement rather than a named provider list, Tables is read-only through the platform tools for the time being but a flow with a Query Node still writes, the agent gets Node-RED validation back and self-corrects, clients supply their own confirmation prompt, and enforcement granularity is teams plus read or write rather than per-tool.
The page said the logs covered FlowFuse Expert and API actions only. A connected agent carries the same sparkle icon, distinguished by the hover label: via Expert for the first party, via MCP for your own agent.
…r tooling The fallback is for the reader whose approved assistant cannot sign in, so that shape goes first. The editor and command-line config, which is the technically simpler one, follows it rather than framing the section.
Is this still the case? I managed to connect on Claude Desktop with OAuth only without what you saying above in the PR body |
Co-authored-by: Andrea Palmieri <76187074+andypalmi@users.noreply.github.com>
andypalmi
left a comment
andypalmi
left a comment
There was a problem hiding this comment.
I'm pretty sure the .md file with line breaks in middle of sentences is going to render badly in the website, have we checked that out?
|
Regarding the line breaks in the middle of sentences: checked it against production rather than guessing. https://flowfuse.com/docs/install/configuration/ already ships hard-wrapped prose, and the live page renders it inside a single Regardless, I have unwrapped the page to one line per paragraph. Every other doc in the repo is written that way and your suggestions all came in as single lines, so no reason to keep it mixed. All suggestions taken and pushed. The only one not changed is the Platform Automation and Flow Building split, which stays on this page for now per that thread. |
andypalmi
left a comment
andypalmi
left a comment
There was a problem hiding this comment.
I'm ok with these changes, just check my latest suggestions and then it's a go from me
Co-authored-by: Andrea Palmieri <76187074+andypalmi@users.noreply.github.com>
3 participants
Adds
docs/user/expert/third-party-agents.mdfor the 3.0 third-party agent feature, plus an entry point from the Expert index.Also updates
docs/user/logs.md, which said the audit log covered Expert and API actions only. A connected agent shows the same sparkle icon, withvia MCPon hover rather thanvia Expert.Endpoint is
https://app.flowfuse.com/mcp, confirmed on the thread. OAuth is the intended route and the page leads with it. Claude Desktop connects with OAuth alone, confirmed on this PR, so the token route is no longer written up as a per-agent workaround. It is a short mention for any client that has no sign-in flow, with no configuration examples and no agent named, pointing at the two common JSON schemas instead.