# eproc > Sequential evaluation library for a stream of bounded scores. Use it after a scorer and before scheduling the next evaluation. Python, CLI and MCP wrap the C11 implementation. Other bindings include WebAssembly, Rust and Go. ## First call Install the Python wheel from the v0.1.0 GitHub release. ```sh python -m pip install --no-index --only-binary=:all: --find-links https://github.com/pgg898200-ship-it/eproc/releases/expanded_assets/v0.1.0 eproc==0.1.0 ``` The [runnable stopping rule](https://github.com/pgg898200-ship-it/eproc#run-a-stopping-rule) feeds a fixed score stream to `EvalGate` and prints `REJECT 41 20.52`. For an upper-tail test, predeclare the bounds, null mean, direction and alpha. The null bounds the conditional mean of every next score given the previous observations. Under that null, the chance of ever crossing 1/alpha is at most alpha. ## Choose an interface - [CLI and Python integration examples](https://github.com/pgg898200-ship-it/eproc#integration-examples). `eproc gate --field score --mu0 0.5 --alpha 0.05 --side upper --stop-early --json` reads a stream of JSON objects with a numeric score field. Inspect `rejected`, `rejected_at`, `e_value` and `n` in the JSON result. - [Connect an MCP client](https://github.com/pgg898200-ship-it/eproc#connect-an-mcp-client). Set the stdio command to the Python interpreter where the release wheel is installed, with args `["-m", "eproc", "mcp"]`. - [Full API reference](https://github.com/pgg898200-ship-it/eproc/blob/main/llms-full.txt). Includes `PairedEvalGate` for paired observations, `ConfidenceSequence` for intervals, `AgentRace` for multi-arm comparisons, and online conformal, forecast fusion and subsampling APIs. - [Single-header C and C++](https://github.com/pgg898200-ship-it/eproc/blob/main/dist/eproc_single.h), [Rust](https://github.com/pgg898200-ship-it/eproc/tree/main/rust/eproc) and [Go](https://github.com/pgg898200-ship-it/eproc/tree/main/go/eproc). - [Local WebAssembly demo](https://pgg898200-ship-it.github.io/eproc/) and [JavaScript bindings](https://github.com/pgg898200-ship-it/eproc/tree/main/js). ## Evidence - [Mathematical assumptions](https://github.com/pgg898200-ship-it/eproc/blob/main/docs/MATHEMATICS.md) - [Fixed-seed simulations and comparisons](https://github.com/pgg898200-ship-it/eproc/blob/main/docs/COMPARISON.md) - [Timing data and workload details](https://github.com/pgg898200-ship-it/eproc/blob/main/benchmarks/BENCHMARKS.md) - [Apache-2.0 license](https://github.com/pgg898200-ship-it/eproc/blob/main/LICENSE)