
Doubleoh Mcp
io.github.cheedliPublished by GitHub account cheedliv0.1.1Updated Oct 24, 2026
The fix desk for AI agents: human fixes become skills the whole fleet reuses.
Overview
DoubleOh's fix loop as an MCP server: agents look up learned skills, request human fixes when stuck, and report whether a skill worked.
- What you can do
- Look up step-by-step procedures learned from past human fixes, with each skill's track record, before retrying a task the agent has failed at before. Request a human fix when stuck; if the fleet already knows the problem, the procedure is returned instantly and nobody is paged, and duplicate requests are answered as already being fixed. Report back whether a followed skill worked, so skills that keep failing stop being served. Refusals such as a private URL or a quota come back as readable tool results rather than protocol errors.
- When to use it
- Useful for agents that repeatedly hit the same walls and could benefit from procedures distilled from human fixes, or that need a way to escalate to a person without spamming a team. Suited to teams running several MCP-capable agents that should share what they learn.
- Requirements
- Runs locally over stdio, typically launched with bunx doubleoh-mcp. Requires a DoubleOh API key in the DOUBLEOH_API_KEY environment variable, issued in the console under Keys and settings. DOUBLEOH_BASE_URL optionally points at a self-hosted deployment and defaults to the hosted service. Network access to the DoubleOh API is needed.
Installation
In SourceWeft
- Open Doubleoh Mcp in the dashboard and add it to a workspace.
- Enable the server for the chats that should use its tools.
Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.
Other MCP clients
Follow the launch instructions in the repository.
README
doubleoh-mcp
The DoubleOh fix loop as an MCP server. Any MCP-capable agent — Claude Code, Claude Desktop, and the growing list of clients — gets the loop as three tools with zero integration code.
The tools
doubleoh_skills_for— before attempting a task it has failed at before, the agent asks what has been learned and receives step-by-step procedures from past human fixes, with each one's track record.doubleoh_request_fix— when stuck, the agent asks for a human. If the fleet already knows the wall, the procedure comes back instantly instead and nobody is paged. Duplicate walls are answered with "already being fixed, N runs blocked" — an agent in a retry loop never spams a team.doubleoh_report_skill— after following a skill, the agent reports whether it worked. Skills that keep failing stop being served.
Refusals (a private URL, a quota) come back as readable tool results, not protocol errors — the model is told what went wrong in a sentence it can act on.
DOUBLEOH_BASE_URL points at a self-hosted deployment; it defaults to the hosted service.
Source: packages/mcp/README.md at commit 1440780
Tools
0Version history
1- v0.1.1LatestSep 29, 2026
Score
40% 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 days25 pts
Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.
- Auth declared10 pts
Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.
- 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
Declare a license in the repository or the registry entry.