FrameOn — Agentic Project Management

com.frameonlabPublished by frameonlab.comOfficial publisherv1.2.1Updated Oct 24, 2026

Project management for AI agents: tasks, docs, decisions and time in one shared team context.

VerifiedStreamable HTTPWeb executableProductivity & WorkflowKnowledge & Memory

Overview

AI-generated overview

FrameOn is a hosted MCP server that gives an AI assistant shared project context — tasks, wiki documents, decisions, project memory, team, alerts and time logs.

What you can do
Read the project state: task tree, wiki where decisions live, separate project memory of conventions and traps, team, alerts and time log, then write back what the agent did so later agents find it. Retrieve a written briefing at frameon://guide explaining which tool answers which question and where the server's limits are. Load skill playbooks through list_skills and get_skill. Of 20 advertised tools, 13 are read-only and 7 write.
When to use it
Use it when an assistant is asked to help with an ongoing team project and needs the real state — earlier decisions, stated constraints, who is already on a task — instead of guessing from pasted chat. Also useful when several agents across machines or clients should share one project context.
Requirements
A remote hosted endpoint over streamable HTTP; nothing is run locally. OAuth 2.1 with Dynamic Client Registration and PKCE (S256) is required: the first call opens a browser authorisation screen where a human approves the workspace, and the client stores and rotates the token. A Personal Access Token sent as an Authorization Bearer header works for scripted use without a browser.
Before you install
Write tools change project data (7 of 20 tools), gated per call on the credential's scope and role. Access is multi-tenant and filtered by tenant, so a credential reaches only the projects its role reaches. A Personal Access Token is a secret to keep out of shared configs. The service implementation, database and tenant layer are closed source and not in the public repository.

Installation

In SourceWeft

  1. Open FrameOn — Agentic Project Management in the dashboard and add it to a workspace.
  2. Connect the required credentials. They are stored privately and never shown on public pages.
  3. Enable the server for the chats that should use its tools.

Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.

Use with an agent

Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.

Read https://sourceweft.com/mcp/com.frameonlab%2Fframeon/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.

Read the install guide the agent follows

Other MCP clients

Add this to your client's mcpServers config. The client completes OAuth sign-in on first connect.

{
  "mcpServers": {
    "frameon-agentic-project-management": {
      "type": "http",
      "url": "https://api.frameonlab.com/api/v1/mcp"
    }
  }
}

README

FrameOn MCP

MCP server for agentic project management. Tasks, documents, decisions and time in one shared team context, reachable by an AI agent over the Model Context Protocol.

This repository holds the public surface of that server: the protocol types, the guide the server serves to connected clients, the skill playbooks, and the published manifest. The server itself is hosted — you do not run it.

Endpointhttps://api.frameonlab.com/api/v1/mcp
Transportstreamable-http
AuthOAuth 2.1 — Dynamic Client Registration, PKCE (S256) required
Tools20 (13 read-only, 7 write) — see docs/tools.md
Registrycom.frameonlab/frameon
Docshttps://app.frameonlab.com/mcp

Why this exists

An AI assistant asked to help with a project normally works from whatever got pasted into the chat. It cannot see what was decided three weeks ago, which constraint someone stated in a meeting, or who is already on the task. So it guesses, confidently, and a human spends the afternoon correcting it.

FrameOn exposes the project itself: the task tree, the wiki where decisions live, a separate project memory holding conventions and traps, the team, the alerts and the time log. The agent reads the real state, writes back what it did, and the next agent — on another machine, in another client — finds it there.

Connect

Claude Code

bash
claude mcp add --transport http frameon https://api.frameonlab.com/api/v1/mcp

Claude Desktop, Cursor, and other mcpServers clients

json
{  "mcpServers": {    "frameon": {      "type": "http",      "url": "https://api.frameonlab.com/api/v1/mcp"    }  }}

ChatGPT

Add a connector pointing at the same URL. The OAuth flow runs in the browser; no key is pasted anywhere.

There is no API key in any of these. The first call opens an authorisation screen, a human approves the workspace, and the client stores a token it rotates on its own. A Personal Access Token also works, as Authorization: Bearer …, for scripted use where no browser exists.

More, including a per-client walkthrough: https://app.frameonlab.com/mcp

What the agent gets

Connect and call frameon://guide. The server hands back a written briefing — what FrameOn is, which tool answers which question, the traps that cost a round-trip, and, just as importantly, what FrameOn has that the agent does not: the Gantt, the approval step on timesheets, the financial reports. A tool that refuses to say where its edges are gets improvised around, badly.

Scope of this repository

PathWhat it is
src/mcp.types.tsJSON-RPC 2.0 envelope types and the protocol versions the server echoes
src/mcp-guide.tsthe text served at frameon://guide, plus the prompt specs
src/mcp-skills.tsthe skill playbooks behind list_skills / get_skill
.mcp.jsondrop-in client config at the repo root — the Open Plugins entry point
server.jsonthe manifest published to the MCP registry
docs/tools.mdthe 20 tools, with the descriptions the server advertises
examples/client configuration, ready to paste

The service implementation, the database and the tenant layer are not here and are not open source. What is here is what a client talks to and what an agent reads — enough to know exactly what you are connecting to before you connect.

Security

The endpoint is multi-tenant and every query filters by tenant. A credential reaches the projects its role reaches and nothing else; tenant_id never appears in a tool response. Write tools are gated per call, on the scope and the role carried by the credential — never on anything sent in the request body.

Found something that looks wrong? [email protected]. Please do not open a public issue for a suspected vulnerability.

Licence

MIT — see LICENSE.

Source: README.md at commit 12e13fe

Tools

0
This server asks you to sign in, so SourceWeft could not read its tools without an account and has not indexed them. What a connected account sees can depend on that account and its permissions.

Version history

3
  1. v1.2.1LatestSep 19, 2026
  2. v1.2.0Sep 19, 2026
  3. v1.0.0Sep 16, 2026

Score

85% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 daysRequired25 pts
  • Auth declaredRequired10 pts
  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts