Find the cheapest AI model that passes your tests: audit repo models, build evals, routing, sizing
MCP Server: co.indigloo/model-finder
co.indigloo/model-finder is an MCP server that helps identify the cheapest AI model that passes your tests. It audits repository models, builds evaluation suites (evals), and supports routing and sizing decisions based on those tests.
🛠️ Key Features
Audit repo models
Build evals
Routing
Sizing
🚀 Use Cases
Compare models from a repository against test results
Select a lower-cost model that still passes evaluation
Determine routing and model sizing using test-driven criteria
⚡ Developer Benefits
Reusable evaluation construction for model comparison
Centralized workflow for auditing, evals, routing, and sizing
⚠️ Limitations
Capabilities described are limited to auditing models, building evals, and applying routing and sizing based on tests.
Find the cheapest AI model that passes your own tests, and keep it that way. Claude · Gemini · OpenAI · Bedrock · Vertex AI · local models — published by Indigloo
Runs locally. Model Finder is a set of four Agent Skills plus a local MCP server (mcp/server.py, Python 3.9+ standard library, no dependencies). The server makes no network calls, needs no API keys, never writes files and never sends anything to Indigloo. Only the eval script and the router it generates call model providers, with your own keys, when you run them.
How it works
flowchart LR
A["Audit<br/><sub>models and SDKs in your repo</sub>"] --> B["Draft evals<br/><sub>from your prompts, PII dropped</sub>"]
B --> C["Evaluate<br/><sub>your keys · pass rate · latency · cost</sub>"]
C --> D["Rank<br/><sub>cost per passing answer</sub>"]
D --> E["Route<br/><sub>cheapest first, escalate on failure</sub>"]
F["Size<br/><sub>weights + KV cache vs your GPU</sub>"] -.-> C
classDef step fill:#f3f0e8,stroke:#1c1b19,color:#1c1b19;
classDef pick fill:#f3f0e8,stroke:#cd462a,color:#cd462a;
class A,B,C,E,F step;
class D pick;
Skill
What it does
find-model
Recommends models across Claude, Gemini, OpenAI, cloud, and local options for your task, budget, and route, including coding subtasks in an IDE.
eval-models
Runs your examples against shortlisted models and measures pass rate, latency, and billed cost with your own keys.
route-models
Builds a cheapest-first routing setup: a standalone router with quality guards plus a LiteLLM config.
size-local-model
Estimates memory for weights and KV cache and compares it with your hardware.
Advisory only: nothing is deployed, your active model is never changed, and data goes only to providers you configure and approve. Personal data and secrets are blocked before sending by default; see DATA-PROTECTION.md for how this maps to India's DPDP Act.
Install
Requires Python 3.9+ on your PATH as python3 (see Windows). Nothing to pip install.
Client
Command
Skills
MCP tools
Claude Code
/plugin marketplace add Abhiindigloo/EvalHarness then /plugin install model-finder@model-finder
✅
✅ via .mcp.json
OpenAI Codex CLI
codex plugin marketplace add Abhiindigloo/EvalHarness then codex plugin add model-finder@model-finder
✅
✅ via mcp.json (Agent Plugins format, runs ./mcp/server.py). Codex gives the server no workspace, so the agent passes the project path
GitHub Copilot CLI
copilot plugin marketplace add Abhiindigloo/EvalHarness then copilot plugin install model-finder@model-finder
Clone the repo once and replace /absolute/path/to/EvalHarness. Repository tools read only inside the workspace: the roots your client shares over MCP, else MODEL_FINDER_ROOT, else the server's working directory. A filesystem root or home directory is refused.
Windows
Codex, Copilot CLI and Cursor start the server as ./mcp/server.py, which relies on its #!/usr/bin/env python3 line; on Windows use the manual setup or uvx model-finder-mcp. If python3 is not on your PATH, the MCP server will not start. Install Python from python.org (tick "Add python.exe to PATH") and either create a python3 alias or use the manual setup with "command": "py" and "args": ["-3", "C:/path/to/EvalHarness/mcp/server.py"].
MCP tools
Ten read-only tools, so the same features work as tools in every supported IDE:
Tool
What it does
modelfinder_audit_models
Lists AI model IDs and SDKs in your repo with file:line locations, routes and hard-coded IDs
modelfinder_draft_eval_examples
Drafts eval examples from prompt files and fixtures, dropping anything with personal data or secrets
Builds and checks an eval config: call count, env var names, unpriced models, guard verdict
modelfinder_check_output
Tests a reply against eval checks or routing guards
modelfinder_rank_eval_results
Ranks results by cost per passing answer, with the cost/reliability cascade
modelfinder_generate_routing
Returns routing.json and a LiteLLM config from eval results
modelfinder_estimate_cost
Monthly token cost and cost per accepted result
modelfinder_estimate_memory
Local memory for weights and KV cache
modelfinder_scan_sensitive_data
Personal-data and secrets scan (categories and counts only)
Every tool returns a readable markdown summary by default and the full result as structured JSON (response_format: "json" puts the JSON in the text too). Large results are paginated (limit, offset, has_more, next_offset), modelfinder_audit_models can filter by provider family or hard-coded IDs, and unknown arguments are rejected with the list of valid ones. The server never opens a network connection, never reads API keys and never writes files. Only eval_models.py and the generated router.py call model providers, with your keys, when you run them.
Quick start
In your IDE, ask: "Audit which AI models this repo uses, then draft evals from our prompts and test whether a cheaper model passes." The assistant uses the MCP tools to audit and draft, agrees checks with you, and asks before the paid run.
From a terminal:
bash
cp skills/eval-models/assets/example-config.json my-eval.json # fill in verified model IDs and pricesexport ANTHROPIC_API_KEY=... GEMINI_API_KEY=... OPENAI_API_KEY=...
python3 skills/eval-models/scripts/repo_scan.py audit . # what the repo uses
python3 skills/eval-models/scripts/repo_scan.py examples . # draft examples from prompt files
python3 skills/eval-models/scripts/eval_models.py my-eval.json --dry-run
python3 skills/eval-models/scripts/eval_models.py my-eval.json --out results
python3 skills/route-models/scripts/make_routing.py --results results/results.json --guards guards.json --out routing
python3 routing/router.py routing/routing.json "your prompt"
End-to-end tests run against a local mock of the OpenAI-compatible, Anthropic, and Gemini APIs, and drive the MCP server over stdio; no key or network access is needed. evaluations/ holds a 10-question agent evaluation of the MCP tools.
Model Finder is free and MIT-licensed. If it saves you money on model calls, you can support its maintenance through GitHub Sponsors. Sponsorship funds maintenance and new features; it does not buy priority support or change what the tools recommend.
Releases
Bump the version everywhere (python3 tests/check_release_version.py lists any field that differs) and add a CHANGELOG entry.
Push a tag vX.Y.Z. release.yml runs the tests, checks the versions, builds the package, publishes it to PyPI with trusted publishing, then publishes server.json to the official MCP Registry.
One-time setup:
PyPI: add a pending trusted publisher for project model-finder-mcp (owner Abhiindigloo, repository EvalHarness, workflow release.yml, environment pypi), ideally from the Indigloo PyPI organization, and create the pypi environment in the repository settings.
MCP Registry (co.indigloo/model-finder, verified through the indigloo.co domain). On your own machine, with OpenSSL 3:
Add the TXT record at the apex of indigloo.co, create a repository environment mcp-registry, store the hex key as its secret MCP_REGISTRY_PRIVATE_KEY, then delete mcp-registry-key.pem or keep it in a password manager.
Repository layout
code
skills/ shared Agent Skills (single source of truth)
mcp/server.py local MCP server (stdio, standard library only)
.mcp.json MCP launch config for Claude Code and Cursor (${CLAUDE_PLUGIN_ROOT}); disabled as a
project config inside this repo by .claude/settings.json
mcp.json Agent Plugins MCP launch config (Codex, Copilot CLI, Cursor): runs ./mcp/server.py
server.json, pyproject.toml MCP Registry entry and PyPI package (model-finder-mcp); pypi/ holds the package marker
glama.json Glama directory ownership
.github/workflows/ tests on every push; release.yml publishes tags to PyPI and the MCP Registry
.claude-plugin/ Claude Code (and Copilot CLI) plugin + marketplace manifests
.codex-plugin/ Codex plugin manifest
.agents/plugins/ Codex repository marketplace
.cursor-plugin/ Cursor plugin + marketplace manifests
gemini-extension.json Gemini CLI extension (context: GEMINI.md)
plugin.json Agent Plugins manifest with OpenAI listing metadata
AGENTS.md, GEMINI.md skill index for agents that read root instructions
PRIVACY.md, TERMS.md, SUPPORT.md pages the marketplace listings link to
assets/ logo, README plates, and DESIGN.md (the visual language)
skills/eval-models/scripts/mf_core.py is duplicated in skills/route-models/scripts/ so each skill works on its own. The test suite fails if the copies differ.