diff --git a/CHANGELOG.md b/CHANGELOG.md
index 8b99b6b..2af61e4 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -8,6 +8,7 @@
 - Added `--skip-git-repo-check` for Codex-backed map, review, fix, and revalidate commands so initialized non-Git roots can run Codex, thanks @im-zayan.
 - Added `CLAWPATCH_CODEX_SANDBOX` for overriding Codex provider sandbox mode when the host already provides isolation, thanks @IAMSamuelRodda.
 - Added `clawpatch review --export-tribunal-ledger` to emit review findings as JSONL for downstream ledger ingestion, thanks @dpdanpittman.
+- Added `clawpatch review --prompt-file` to append extra reviewer guidance from a file or stdin, thanks @dpdanpittman.
 - Added deterministic Express, Fastify, and Hono route mapping for Node projects, thanks @rohitjavvadi.
 - Fixed provider commands with relative `--root` paths by canonicalizing explicit roots before invoking Codex or other providers.
 - Added first-pass Elixir Mix/Phoenix mapping for project metadata, contexts, Phoenix web slices, runtime config, Ecto migrations, project scripts, ExUnit tests, and Mix validation defaults, thanks @tears-mysthrala.
diff --git a/src/app.ts b/src/app.ts
index a11baec..365d237 100644
--- a/src/app.ts
+++ b/src/app.ts
@@ -1,4 +1,4 @@
-import { writeFile } from "node:fs/promises";
+import { readFile, writeFile } from "node:fs/promises";
 import { join, resolve } from "node:path";
 import { hostname } from "node:os";
 import {
@@ -239,6 +239,7 @@ export async function reviewCommand(
   const config = applyProviderFlags(loaded.config, flags);
   const provider = providerByName(config.provider.name);
   const mode = reviewMode(flags);
+  const customPrompt = await loadCustomReviewPrompt(flags);
   const features = await selectReviewFeatures(loaded, flags);
   if (features.length === 0 && typeof flags["since"] === "string") {
     if (flags["dryRun"] === true) {
@@ -303,6 +304,7 @@ export async function reviewCommand(
             index,
             total: features.length,
             mode,
+            customPrompt,
             allowNonPendingFeatureReview: stringFlag(flags, "feature") !== undefined,
           });
           findingIds.push(...reviewed.findingIds);
@@ -594,6 +596,7 @@ type ReviewFeatureOptions = {
   index: number;
   total: number;
   mode: ReviewMode;
+  customPrompt: string | null;
   allowNonPendingFeatureReview: boolean;
 };
 
@@ -608,6 +611,7 @@ async function reviewFeature(options: ReviewFeatureOptions): Promise<{ findingId
     index,
     total,
     mode,
+    customPrompt,
     allowNonPendingFeatureReview,
   } = options;
   const started = Date.now();
@@ -634,6 +638,7 @@ async function reviewFeature(options: ReviewFeatureOptions): Promise<{ findingId
       lockedFeature,
       config,
       mode,
+      customPrompt,
     );
     const output = await provider.review(loaded.root, prompt, providerOptions(config));
     const modeFindings = reviewFindingsForMode(output.findings, mode);
@@ -1201,6 +1206,39 @@ function reviewMode(flags: Record<string, string | boolean>): ReviewMode {
   throw new ClawpatchError("invalid --mode; expected default or deslopify", 2, "invalid-usage");
 }
 
+async function loadCustomReviewPrompt(
+  flags: Record<string, string | boolean>,
+): Promise<string | null> {
+  const path = stringFlag(flags, "promptFile");
+  if (path === undefined) {
+    return null;
+  }
+  if (path === "" || path === "-") {
+    return readStdinToString();
+  }
+  try {
+    return await readFile(resolve(path), "utf8");
+  } catch (error: unknown) {
+    const message = error instanceof Error ? error.message : String(error);
+    throw new ClawpatchError(
+      `failed to read --prompt-file ${path}: ${message}`,
+      2,
+      "invalid-usage",
+    );
+  }
+}
+
+async function readStdinToString(): Promise<string> {
+  if (process.stdin.isTTY) {
+    throw new ClawpatchError("--prompt-file=- requested but stdin is a TTY", 2, "invalid-usage");
+  }
+  const chunks: Buffer[] = [];
+  for await (const chunk of process.stdin) {
+    chunks.push(Buffer.isBuffer(chunk) ? chunk : Buffer.from(chunk));
+  }
+  return Buffer.concat(chunks).toString("utf8");
+}
+
 function reviewFindingsForMode(
   findings: ReviewOutput["findings"],
   mode: ReviewMode,
diff --git a/src/cli.ts b/src/cli.ts
index e49044a..894918d 100644
--- a/src/cli.ts
+++ b/src/cli.ts
@@ -160,6 +160,7 @@ const commandFlags = {
     "reasoningEffort",
     "skipGitRepoCheck",
     "dryRun",
+    "promptFile",
     "exportTribunalLedger",
   ]),
   report: new Set(["status", "severity", "feature", "project", "category", "triage", "output"]),
@@ -206,6 +207,7 @@ const valueFlagNames = new Set([
   "provider",
   "model",
   "reasoning-effort",
+  "prompt-file",
   "export-tribunal-ledger",
   "output",
   "status",
@@ -391,6 +393,8 @@ Flags:
   --reasoning-effort <none|minimal|low|medium|high|xhigh>
   --skip-git-repo-check
   --dry-run
+  --prompt-file <path>    appends extra reviewer guidance to the prompt;
+                          use "-" to read from stdin
   --export-tribunal-ledger <path>
                           after the review completes, emit a single
                           JSONL file with one line per finding shaped
diff --git a/src/prompt.ts b/src/prompt.ts
index 6c8b68d..6831791 100644
--- a/src/prompt.ts
+++ b/src/prompt.ts
@@ -60,6 +60,7 @@ export async function buildReviewPrompt(
   feature: FeatureRecord,
   config: ClawpatchConfig,
   mode: ReviewMode = "default",
+  customPrompt: string | null = null,
 ): Promise<string> {
   const owned = feature.ownedFiles.slice(0, config.review.maxOwnedFiles);
   const context = feature.contextFiles.slice(0, config.review.maxContextFiles);
@@ -67,6 +68,14 @@ export async function buildReviewPrompt(
   for (const ref of [...owned, ...context]) {
     fileBlocks.push(await fileBlock(root, ref.path));
   }
+  const customBlock =
+    customPrompt !== null && customPrompt.trim() !== ""
+      ? `Additional reviewer guidance (provided via --prompt-file):
+
+${customPrompt.trim()}
+
+`
+      : "";
   return `You are reviewing one semantic feature for clawpatch.
 
 Return strict JSON only. No markdown fences.
@@ -77,7 +86,7 @@ ${JSON.stringify({ name: project.name, detected: project.detected }, null, 2)}
 Feature:
 ${JSON.stringify(feature, null, 2)}
 
-Review categories:
+${customBlock}Review categories:
 - correctness bugs
 - security issues
 - race/concurrency bugs
diff --git a/src/workflow.test.ts b/src/workflow.test.ts
index ed80fdf..3130454 100644
--- a/src/workflow.test.ts
+++ b/src/workflow.test.ts
@@ -2601,6 +2601,124 @@ describe("workflow", () => {
     expect(prompt).toContain("do not report correctness, security, API contract");
   });
 
+  it("injects --prompt-file content into the review prompt", async () => {
+    const root = await fixtureRoot("clawpatch-prompt-file-");
+    await writeFixture(root, "package.json", JSON.stringify({ name: "prompt-file" }));
+    await writeFixture(root, "src/index.ts", "export function main() { return 1; }\n");
+    const context = await makeContext(testOptions(root));
+
+    await initCommand(context, {});
+    const project = await readProject(statePaths(join(root, ".clawpatch")));
+    expect(project).toBeDefined();
+    const promptWithCustom = await buildReviewPrompt(
+      root,
+      project!,
+      {
+        schemaVersion: 1,
+        featureId: "feat_prompt_file",
+        title: "prompt-file",
+        summary: "prompt-file",
+        kind: "library",
+        source: "test",
+        confidence: "high",
+        entrypoints: [{ path: "src/index.ts", symbol: null, route: null, command: null }],
+        ownedFiles: [{ path: "src/index.ts", reason: "test" }],
+        contextFiles: [],
+        tests: [],
+        tags: [],
+        trustBoundaries: [],
+        status: "pending",
+        lock: null,
+        findingIds: [],
+        patchAttemptIds: [],
+        analysisHistory: [],
+        createdAt: new Date().toISOString(),
+        updatedAt: new Date().toISOString(),
+      },
+      await loadConfig(root, testOptions(root)),
+      "default",
+      "Focus exclusively on race conditions and lock ordering bugs.",
+    );
+
+    expect(promptWithCustom).toContain(
+      "Additional reviewer guidance (provided via --prompt-file):",
+    );
+    expect(promptWithCustom).toContain(
+      "Focus exclusively on race conditions and lock ordering bugs.",
+    );
+    // Custom guidance must land before the JSON shape and file blocks so
+    // the model reads it as setup, not as part of the response template.
+    const guidanceIdx = promptWithCustom.indexOf("Additional reviewer guidance");
+    const jsonIdx = promptWithCustom.indexOf("JSON shape:");
+    expect(guidanceIdx).toBeGreaterThan(0);
+    expect(guidanceIdx).toBeLessThan(jsonIdx);
+  });
+
+  it("leaves the review prompt unchanged when --prompt-file is omitted", async () => {
+    const root = await fixtureRoot("clawpatch-prompt-file-omit-");
+    await writeFixture(root, "package.json", JSON.stringify({ name: "prompt-file-omit" }));
+    await writeFixture(root, "src/index.ts", "export function main() { return 1; }\n");
+    const context = await makeContext(testOptions(root));
+
+    await initCommand(context, {});
+    const project = await readProject(statePaths(join(root, ".clawpatch")));
+    expect(project).toBeDefined();
+    const baseline = await buildReviewPrompt(
+      root,
+      project!,
+      {
+        schemaVersion: 1,
+        featureId: "feat_prompt_file_omit",
+        title: "prompt-file-omit",
+        summary: "prompt-file-omit",
+        kind: "library",
+        source: "test",
+        confidence: "high",
+        entrypoints: [{ path: "src/index.ts", symbol: null, route: null, command: null }],
+        ownedFiles: [{ path: "src/index.ts", reason: "test" }],
+        contextFiles: [],
+        tests: [],
+        tags: [],
+        trustBoundaries: [],
+        status: "pending",
+        lock: null,
+        findingIds: [],
+        patchAttemptIds: [],
+        analysisHistory: [],
+        createdAt: new Date().toISOString(),
+        updatedAt: new Date().toISOString(),
+      },
+      await loadConfig(root, testOptions(root)),
+    );
+
+    expect(baseline).not.toContain("Additional reviewer guidance");
+  });
+
+  it("parses --prompt-file as a review value flag", () => {
+    expect(parseArgs(["review", "--prompt-file", "/tmp/foo.md"]).flags).toMatchObject({
+      promptFile: "/tmp/foo.md",
+    });
+  });
+
+  it("runs review --prompt-file through the CLI entrypoint", async () => {
+    const root = await fixtureRoot("clawpatch-prompt-file-cli-");
+    await writeFixture(root, "package.json", JSON.stringify({ name: "prompt-file-cli" }));
+
+    await runCli(["--root", root, "--json", "--quiet", "init"]);
+
+    await expect(
+      runCli([
+        "--root",
+        root,
+        "--json",
+        "--quiet",
+        "review",
+        "--prompt-file",
+        join(root, "missing.md"),
+      ]),
+    ).rejects.toThrow("failed to read --prompt-file");
+  });
+
   it("writes a tribunal-shaped JSONL ledger when --export-tribunal-ledger is set", async () => {
     const root = await fixtureRoot("clawpatch-export-tribunal-");
     await writeFixture(