proxy
Local-first LLM proxy that meters what every agent run costs and kills runaways before they drain your budget. Free, MIT.
Runs with mockssource: GitHubhomepageTypeScriptMITcommit bdceda091cbc
TypeScript, MIT licensed. The project labels itself: agent cost, claude code, cost attribution, llm gateway, llm proxy, local first, mcp and observability.
proxy runs, with stand-ins for the services it depends on. An Argusic agent installed it in 20 minutes and hit 5 errors and fixed every one without help on a clean machine with no GPU, and the whole session was recorded.
RelayPlane proxy v1.9.64 builds from source, starts on port 4100/4101/4102, answers HTTP 200 on /health, CLI --help works, and 125 of 138 test files pass (1769 of 1832 tests). The 13 failing files are content/marketing-site/CI-workflow/test suites that require files from a monorepo structure not present in this standalone checkout.
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 92 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.
- nanoid v5 is ESM-only but @relayplane/core and other packages use require('nanoid') (CJS)5 minutes
- Missing 'default' export in @relayplane/* package.json exports3 minutes
- vitest.config.ts alias points to ../learning-engine/src/index.ts which doesn't exist2 minutes
- crypto global not available in Node 18 (crypto.randomUUID used in background-detector.test.ts)2 minutes
- tsx binary not in .bin directory (launcher.test.ts)1 minute
- Install time
- 20 minutes
- Cold machine to finish
- 19 minutes
- Errors hit and fixed
- 5 hit, 5 fixed with no human help
- How the result was proved
- curl -s -o /dev/null -w '%{http_code}' http://localhost:4102/health returned 200; npx vitest run: 125 passed, 60 failed (all failures are marketing-site/doc-content tests referencing files outside this standalone repo)
- Model tokens used
- 156,743
- Exact commit tested
- bdceda091cbc
- 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.
RelayPlane Proxy - Intelligent AI Model Routing === SERVER (timeout 3s) === ╭─────────────────────────────────────────╮ │ RelayPlane Proxy v1.9.64 │ │ Intelligent AI Model Routing │ ╰─────────────────────────────────────────╯ === TESTS === Start at 00:50:56 Duration 52.53s (transform 1.46s, setup 3ms, collect 3.44s, tests 48.49s, environment 0ms, prep… 2026-10-31T00:52:50.376742Z ERROR codex_core::session: failed to record rollout items: thread 01a18… tokens used 156,743
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
- Installed in 20 minutes.
- Recovered from all 5 errors without a human stepping in, which says the failures are documented well enough to solve.
- Licensed MIT, as reported by its host.
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.
- Measured once so far. A single attempt shows it can run, not that it runs reliably.
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 | 1/3 | Runs with mocks | 92.00 | 0.45 |
Topics (from GitHub)
agent-costclaude-codecost-attributionllm-gatewayllm-proxylocal-firstmcpobservabilityopen-source
Embed the badge
Markdown for the project README. It links back here; terms on the terms page.
[](https://argusic.com/subject/proxy)Questions
- Does proxy run?
- proxy runs, with mocks standing in for external services. Argusic installed and launched it on a clean machine in 20 minutes, hitting 5 errors on the way, and recorded the session.
- How did Argusic test proxy?
- On a fresh, disposable machine, with every command recorded and the repository pinned at commit bdceda091cbc. One attempt is recorded, and the full method is on the methodology page.
- What did it cost to test proxy?
- The run that produced this verdict cost $0.45: real compute and model cost, not a list price per million tokens. It is recorded on that run's page.
- How long does proxy take to install?
- 20 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 proxy need a GPU?
- Not to start. Argusic reached a running state on a machine with no GPU.
- What goes wrong when installing proxy?
- 5 things broke in the recorded run, and 5 were fixed without human help. Each one, and the time it cost, is listed on this page.
- Where is the evidence for proxy?
- The recorded run is on this page, each linking to its full log and terminal recording, stored with a sha256 fingerprint so it cannot be quietly altered.