diff --git a/README.md b/README.md
index fac402f3..2aec1acc 100644
--- a/README.md
+++ b/README.md
@@ -348,6 +348,7 @@ The [independent model research](community/projects/tools/README.md#independent-
 - [jev-mcp](https://github.com/jkudish/jev-mcp) - MCP server with ten purpose-built TypeSafe Jev judgment tools (verify, screen, find, classify, review, gate, and more). [Project guide](community/projects/tools/jev-mcp.md).
 - [jev-mcp-server](https://github.com/wangkuangkuang/jev-mcp-server) - Python MCP server for official TypeSafe Jev choice/score/noul plus compare/verify/classify and a one-command client installer. [Project guide](community/projects/tools/jev-mcp-server.md).
 - [jev2mcp](https://github.com/n8mirai/jev2mcp) - Local companion + Chrome extension: TypeSafe Jev picks ChatGPT MCP/plugin/tool mentions from a configured catalog. [Project guide](community/projects/tools/jev2mcp.md).
+- [jevper](https://github.com/zhulinchng/jevper) - Jev-shaped System One noul/choice/score client over any OpenAI-compatible model (no hosted TypeSafe API). [Project guide](community/projects/tools/jevper.md).
 - [Jevs](https://github.com/nshcr/jevs) - Bun/TypeScript MCP server and Codex plugin for TypeSafe Jev classify/score/check/batch via the official JS SDK. [Project guide](community/projects/tools/jevs.md).
 - [Jev MCP (Freepik)](https://github.com/freepik-company/jev-mcp) - Go MCP server for typed decide/classify/verify/rerank with Jev via OpenRouter or TypeSafe (binary/container releases). [Project guide](community/projects/tools/freepik-jev-mcp.md).
 - [tokengate (jev-model-tokengate)](https://github.com/Thanh-Mathieu95/jev-model-tokengate) - Streaming OpenAI-compatible proxy that buffers tokens and runs TypeSafe Jev safety criteria before release, aiming for zero UI leakage vs post-hoc redaction. [Project guide](community/projects/tools/jev-model-tokengate.md).
diff --git a/community/projects/tools/README.md b/community/projects/tools/README.md
index 70e8f1ee..02b3bb96 100644
--- a/community/projects/tools/README.md
+++ b/community/projects/tools/README.md
@@ -338,6 +338,7 @@ These projects study related typed-decision patterns using other models. They ar
 | [jev-mcp](jev-mcp.md) | Give agents ten purpose-built TypeSafe Jev judgment MCP tools (verify, screen, find, classify, review, gate, …). | TypeScript · npm MCP server (`@jkudish/jev-mcp`) |
 | [jev-mcp-server](jev-mcp-server.md) | MCP for official TypeSafe Jev choice/score/noul plus compare/verify/batch classify and client installer. | Python · PyPI MCP (`jev-mcp-server` 0.2.3) |
 | [jev2mcp](jev2mcp.md) | Local companion + Chrome extension: TypeSafe Jev selects ChatGPT MCP/plugin/tool mentions from your catalog. | Node.js · local server + extension (`jev2mcp` 0.2.1) |
+| [jevper](jevper.md) | Jev-shaped System One noul/choice/score over any OpenAI-compatible client (no hosted TypeSafe API). | Python · PyPI (`jevper` 0.1.2, Apache-2.0) |
 | [Jevs](jevs.md) | Call TypeSafe Jev classify/score/check/batch from Bun MCP / Codex plugin via the official JS SDK. | TypeScript · Bun MCP / Codex plugin (`jevs` 0.1.0) |
 | [Jev MCP (Freepik)](freepik-jev-mcp.md) | Go MCP server for typed decide/classify/verify/rerank via OpenRouter or TypeSafe (binary/container). | Go · MCP binary (`jev-mcp` v0.3.0) |
 | [Jev Studio](jev-studio.md) | Experiment with TypeSafe Jev via a `jev` CLI (verify/screen/classify/…) and an MCP server with cookbook tools. | Python · CLI + MCP (`jev-studio` 0.1.0 Alpha) |
diff --git a/community/projects/tools/jevper.md b/community/projects/tools/jevper.md
new file mode 100644
index 00000000..353fb05f
--- /dev/null
+++ b/community/projects/tools/jevper.md
@@ -0,0 +1,49 @@
+# jevper
+
+[All projects](../README.md) · [SDKs and integrations](README.md#sdks-and-integrations)
+
+Jev-shaped System One client over any OpenAI-compatible model: `noul` / `choice` / `score` with probabilities and confidence—without calling the hosted TypeSafe API. Independent of `typesafe-sdk`.
+
+| At a glance | Details |
+| --- | --- |
+| Source | [Source](https://github.com/zhulinchng/jevper) |
+| Maintainer | [zhulinchng](https://github.com/zhulinchng). Independently curated; this entry is not an upstream submission or endorsement. Not affiliated with TypeSafe AI. |
+| Format | Python package **`jevper` 0.1.2** on PyPI (`pydantic` only at runtime; duck-typed OpenAI-like client). |
+| Requirements | Python ≥ 3.10; an object exposing `responses.create` or `chat.completions.create` (e.g. OpenAI SDK or llama.cpp server). |
+| License | [Apache-2.0](https://github.com/zhulinchng/jevper/blob/c95c140cc7604f2bfb97064030d06d61ab0c9298/LICENSE). |
+| Disclosure | AI-assisted catalog review; no affiliation. Listing is not an endorsement. Source inspected (README, LICENSE, pyproject). Live LLM calls were **not** executed on the review host. Behavior is not validated against official Jev. |
+
+## When to use
+
+Use it when application code expects System One-shaped answers but you want to back them with your own OpenAI-compatible model. Prefer the official TypeSafe SDK/API for hosted Jev; prefer [nokia AnyJev](nokia-anyjev.md) for local logit/head research on open weights.
+
+## How it works
+
+`SystemOneClient.system_one(state, questions=…)` builds messages and uses `logprobs` (or structured-output methods documented upstream) to fill Choice/Noul/Score answers with probabilities and confidence. No dependency on `typesafe-sdk` or `openai` at runtime beyond the duck-typed client you pass in.
+
+## Get started
+
+```sh
+pip install jevper
+# or:
+git clone https://github.com/zhulinchng/jevper.git
+cd jevper
+git checkout c95c140cc7604f2bfb97064030d06d61ab0c9298
+uv venv && uv pip install -e '.[test]'
+pytest -q
+```
+
+## Examples and demos
+
+- README `Choice` example with `SystemOneClient`.
+- Test suite under the repo (offline where fixtures allow).
+
+## Limits and data handling
+
+Requests go to whatever backend your client targets—not TypeSafe. Calibration and option-order robustness depend on that model; this is not official Jev. This listing did not call live models.
+
+## Review and maintenance
+
+Reviewed on **2026-09-23** at [commit c95c140](https://github.com/zhulinchng/jevper/tree/c95c140cc7604f2bfb97064030d06d61ab0c9298) (**0.1.2**, Apache-2.0). AI-assisted source review of README, LICENSE, pyproject. No live model spend.
+
+Related: [AnyJev (Nokia Applied Research)](nokia-anyjev.md), [hunch](hunch.md), [Advocaat](advocaat.md).