ref-tools-mcp
Helping coding agents never make mistakes working with public or private libraries without wasting the context window.
Runs with mockssource: GitHubhomepageTypeScriptMITcommit 83970a356277
TypeScript, MIT licensed. The project labels itself: coding agent, documentation tool, mcp and mcp server.
ref-tools-mcp runs, with stand-ins for the services it depends on. An Argusic agent installed it in 12 minutes and hit 3 errors and fixed every one without help on a clean machine with no GPU, and the whole session was recorded.
After npm install, npm run build, and npm run check all succeeded, the app builds to an executable dist/index.cjs; the HTTP server answered 200 'pong' on /ping and a full MCP session (initialize, tools/list, tools/call search/read, prompts) returned 200 with correct results against a mock Ref API, while the stdio transport answered initialize with serverInfo Ref 3.0.3.
What the agent ended up with on a clean machine, in its own words. How this is measured
Measured by Argusic on a fresh machine every time. Every number links to its evidence. Argusic Score 56 of 100 (the mean of this project's run scores).
At a glance
Subject data from GitHub, linked at the top of this page, refreshed . Test data by Argusic (CC BY 4.0); every number links to a run page with the full log, the recording, and their sha256 hashes.
What was measured
From the run that decided this verdict. Every line is a recorded value, not an estimate.

What went wrong, in the agent's own words
The agent's own words, worst first, never rewritten into advice.
- Background mock/server processes terminated when their launching shell session ended4 minutes
- Python urllib connected to localhost over IPv6, getting EADDRNOTAVAIL against the IPv4 HTTP listener2 minutes
- Test client crashed on MCP notifications/initialized empty 202 response1 minute
- Install time
- 12 minutes
- Cold machine to finish
- 3 minutes
- Errors hit and fixed
- 3 hit, 3 fixed with no human help
- How the result was proved
- curl http://127.0.0.1:8099/ping returned 200 with body 'pong'; npm run build and npm run check exited 0; a full MCP session over HTTP (initialize, tools/list, tools/call ref_search_documentation, tools/call ref_read_url, prompts/list, prompts/get) returned HTTP 200 with correct JSON-RPC results; stdio initialize on node dist/index.cjs returned serverInfo Ref 3.0.3
- Model tokens used
- 49,590
- Exact commit tested
- 83970a356277
- Container image
- sha256:cdd920bce783
What was not measured
The other half, so you can see exactly how far this evidence goes. A gap is printed as a gap and never filled in.
- Peak GPU memory
- this ran on a CPU machine, so there was no GPU to sample
- Behaviour over a long run
- each attempt uses the smallest documented settings and a fixed time budget, so this measures whether it runs, not how it holds up under load
- Behaviour against real services
- this run was verified against stand-in services only
The end of the recording
The last lines of the terminal session that decided this verdict, exactly as they were recorded.
- Read the README: this is `ref-tools-mcp`, an MCP server (stdio by default, or HTTP via `TRANSPORT… - Ran a standard install: `npm install` succeeded (249 packages; Node 18 vs the inspector's `>=22.7… - Built: `npm run build` → `dist/index.cjs` (1.4 mb, executable `755`). Type-check: `npm run check`… - Launched HTTP server (`TRANSPORT=http PORT=8099`): printed `Ref MCP Server running on HTTP at htt… - Because real Ref API credentials would be required for the actual proxied calls, I stood up a sch… - Verified HTTP transport end-to-end: `initialize` → 200 + session id; `tools/list` → `ref_search_d… - Verified stdio transport: a `node dist/index.cjs` child responded on stdout to `initialize` with … **Errors encountered (all fixed by me)** - Background service processes died when their launching shell session ended — relaunched everythin… - Python `urllib` resolving `localhost` to IPv6 (`EADDRNOTAVAIL`) against the IPv4 listener — switc… - My test client crashed on the MCP `notifications/initialized` 202/empty response — made the parse… **Result JSON**
Replay the whole session, every command from a clean machine to this point.
How it was tested
One agent, one container, no human help. It installs the project the way its own documentation says to, runs it, and fixes what breaks. Everything below is recorded as it happened: the terminal session, the log and the exact commit. The full method.
Strengths and limits
Measured facts, not opinions. How this is written.
What went well
- Recovered from all 3 errors without a human stepping in, which says the failures are documented well enough to solve.
- Licensed MIT, as reported by its host.
- Measured 2 times, so the result is not a one-off.
What did not, or is not known
- Only reached a running state against stand-in services, so nothing here proves it works against the real ones.
- Took 12 minutes to install, slower than the median of the 5 comparable projects Argusic has measured.
Of the 6 coding-agent projects Argusic has installed and timed, ref-tools-mcp was the 5th fastest to reach a running state, and 1 of 6 reached one at all.
Also tested, in the same area
Every one of these was installed and run by Argusic on a clean machine. Nothing appears here that was not tested.
Run history
| tested by | attempt | outcome | score | cost USD | when |
|---|---|---|---|---|---|
| Argusic Runner | 2/3 | Did not run | 20.00 | 0.00 | |
| Argusic Runner | 1/3 | Runs with mocks | 92.00 | 0.03 |
Topics (from GitHub)
Embed the badge
Markdown for the project README. It links back here; terms on the terms page.
[](https://argusic.com/subject/ref-tools-mcp)Questions
- Does ref-tools-mcp run?
- ref-tools-mcp runs, with mocks standing in for external services. Argusic installed and launched it on a clean machine in 12 minutes, hitting 3 errors on the way, and recorded the session.
- How did Argusic test ref-tools-mcp?
- On a fresh, disposable machine, with every command recorded and the repository pinned at commit 83970a356277. 2 attempts are recorded, and the full method is on the methodology page.
- What did it cost to test ref-tools-mcp?
- The run that produced this verdict cost $0.03: real compute and model cost, not a list price per million tokens. It is recorded on that run's page.
- How long does ref-tools-mcp take to install?
- 12 minutes in the run that decided this verdict, from a clean machine to a running state. That is one measured attempt, not an average.
- Does ref-tools-mcp need a GPU?
- Not to start. Argusic reached a running state on a machine with no GPU.
- What goes wrong when installing ref-tools-mcp?
- 3 things broke in the recorded run, and 3 were fixed without human help. Each one, and the time it cost, is listed on this page.
- How does ref-tools-mcp compare with the alternatives?
- Of the 6 coding-agent projects Argusic has installed and timed, ref-tools-mcp was the 5th fastest to reach a running state, and 1 of 6 reached one at all.
- Where is the evidence for ref-tools-mcp?
- All 2 recorded runs are on this page, each linking to its full log and terminal recording, stored with a sha256 fingerprint so it cannot be quietly altered.