From 994e906a9f1fe3a00b8668bfc8aa6c5b349a022c Mon Sep 17 00:00:00 2001
From: Postil Maintainer <morgaesis+git@morgaes.is>
Date: Wed, 12 Aug 2026 00:56:17 +0000
Subject: [PATCH] Guard benchmark baseline authority

---
 Cargo.lock                         |  2 +-
 Cargo.toml                         |  2 +-
 bench/baseline.json                |  2 +-
 bench/src/compare-baseline.test.ts | 20 ++++++++++++++++++++
 bench/src/live.ts                  |  2 +-
 5 files changed, 24 insertions(+), 4 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index 064f566..cbf060a 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -1104,7 +1104,7 @@ checksum = "a89322df9ebe1c1578d689c92318e070967d1042b512afbe49518723f4e6d5cd"
 
 [[package]]
 name = "postil-cli"
-version = "0.8.9"
+version = "0.8.10"
 dependencies = [
  "anyhow",
  "assert_cmd",
diff --git a/Cargo.toml b/Cargo.toml
index 1f1772b..7e54932 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "postil-cli"
-version = "0.8.9"
+version = "0.8.10"
 edition = "2024"
 description = "Postil: a low-noise AI review gate. Silent on clean PRs, hard gate on real risk."
 license = "Apache-2.0"
diff --git a/bench/baseline.json b/bench/baseline.json
index 9abd690..dfcaf7a 100644
--- a/bench/baseline.json
+++ b/bench/baseline.json
@@ -2,7 +2,7 @@
   "schemaVersion": 1,
   "corpus": {
     "fixtureCorpusSha256": "3c4f952f9411bf2c4e5f0a61b5975e47d203f7681a8235d53d717c743ea9cb54",
-    "evaluatorSha256": "64dd1975bd5743e15b073b8c6dbad91e37cef3388f793073b216d002de80d78a"
+    "evaluatorSha256": "b0d452b0c2f8d419ef6f4780791cde8cd760d318a82b439aa5746a226305e125"
   },
   "profiles": {
     "z-ai/glm-5.2": {
diff --git a/bench/src/compare-baseline.test.ts b/bench/src/compare-baseline.test.ts
index 5ab4a32..9a7cf72 100644
--- a/bench/src/compare-baseline.test.ts
+++ b/bench/src/compare-baseline.test.ts
@@ -1,4 +1,8 @@
 import { describe, expect, test } from "bun:test";
+import { createHash } from "node:crypto";
+import { readFile } from "node:fs/promises";
+import { resolve } from "node:path";
+import { cases } from "../fixtures/cases";
 import {
   DETECTION_RATE_MAX_DROP_PP,
   FALSE_FINDINGS_MAX_INCREASE,
@@ -12,6 +16,22 @@ import {
   type BaselineProfile,
   type LiveReportForComparison,
 } from "./compare-baseline";
+import { benchmarkCase } from "./harness";
+import { evaluatorSourceSha256 } from "./live";
+
+test("committed baseline authority matches the current benchmark sources", async () => {
+  const baseline = JSON.parse(
+    await readFile(resolve(import.meta.dir, "..", "baseline.json"), "utf8"),
+  ) as {
+    corpus: { fixtureCorpusSha256: string; evaluatorSha256: string };
+  };
+  const fixtureCorpusSha256 = createHash("sha256")
+    .update(JSON.stringify(cases.map((input) => benchmarkCase.parse(input))))
+    .digest("hex");
+
+  expect(baseline.corpus.fixtureCorpusSha256).toBe(fixtureCorpusSha256);
+  expect(baseline.corpus.evaluatorSha256).toBe(await evaluatorSourceSha256());
+});
 
 function fakeReport(overrides: {
   falsePositives?: number;
diff --git a/bench/src/live.ts b/bench/src/live.ts
index 5437d78..c4a632c 100644
--- a/bench/src/live.ts
+++ b/bench/src/live.ts
@@ -380,7 +380,7 @@ async function screeningProfileMetadata(path: string): Promise<{
   };
 }
 
-async function evaluatorSourceSha256(): Promise<string> {
+export async function evaluatorSourceSha256(): Promise<string> {
   const benchRoot = resolve(fileURLToPath(import.meta.url), "..", "..");
   const sources = [
     "fixtures/cases.ts",