diff --git a/CHANGELOG.md b/CHANGELOG.md
index 7ec6aedb60..acbf704144 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -36,6 +36,8 @@
 
 ### 🔧 Bug Fixes
 
+- **fix(github):** drop a trailing assistant prefill before dispatching to GitHub Copilot chat to avoid 400 errors. (thanks @baslr)
+
 - **providers (Kiro — Claude Sonnet 5):** the Kiro provider's model catalog was missing `claude-sonnet-5`, so the model could not be selected or routed even on accounts that already had access to it ("claude-sonnet-5 is not supported"). Added the model to the Kiro registry (`open-sse/config/providers/registry/kiro/index.ts`) as a 1M-context / 128K-output Claude model, mirroring the existing Claude entries; the registry `models[]` feeds both the model selector and the live CodeWhisperer `ListAvailableModels` fallback, so the model is now selectable and routable. Regression guard: `tests/unit/kiro-claude-sonnet-5-2267.test.ts`. (thanks [@openbioinfo](https://github.com/openbioinfo))
 
 - **settings (model aliases — self-heal after restart):** the Settings → Routing page showed "No exact-match aliases configured" after a server restart even though the aliases were persisted in the DB. Aliases are held in a module-local `_customAliases` map in `modelDeprecation.ts` that the boot path hydrates, but Next.js compiles the app-route module graph separately from the startup graph (the same webpack chunk-splitting class as #5312), so the `GET /api/settings/model-aliases` handler read a different, un-hydrated copy. The handler now self-heals: when its in-memory alias map is empty it reads `settings.modelAliases` from the DB (via the existing `getSettings()` db module — no raw SQL in the route) and repopulates the map, so the UI reflects the persisted aliases on the first GET after a restart. Follow-up: the root cause is now also fixed — the `_customAliases` store in `modelDeprecation.ts` is backed by `globalThis` (key `__omniroute_customAliases__`), so the startup and app-route module graphs share **one** store and the route reads the boot-hydrated aliases directly (the DB self-heal remains as a harmless fallback), mirroring the same `globalThis` singleton pattern already applied to `thinkingBudget.ts`/`backgroundTaskDetector.ts` (#5312). Regression guards: `tests/unit/model-aliases-settings-route-selfheal.test.ts` + `tests/unit/model-aliases-globalthis-5777.test.ts`. ([#5777](https://github.com/diegosouzapw/OmniRoute/pull/5777) — thanks [@jleonar2](https://github.com/jleonar2))
diff --git a/open-sse/executors/github.ts b/open-sse/executors/github.ts
index f2507bd9cf..5271500d31 100644
--- a/open-sse/executors/github.ts
+++ b/open-sse/executors/github.ts
@@ -149,6 +149,18 @@ export class GithubExecutor extends BaseExecutor {
       );
     }
 
+    // GitHub Copilot's /chat/completions endpoint rejects a conversation that ends
+    // with an assistant message: "This model does not support assistant message
+    // prefill. The conversation must end with a user message." (HTTP 400). Anthropic
+    // clients such as newest Claude Desktop send a trailing assistant turn as a
+    // prefill seed — the Anthropic API honors it, but Copilot does not. Drop it here,
+    // scoped to the GitHub executor only (the shared translator/contextManager and
+    // other providers that DO honor prefill are untouched).
+    // Port of 9router#2143 (author: Manuel <baslr@users.noreply.github.com>).
+    if (Array.isArray(modifiedBody.messages)) {
+      modifiedBody.messages = this.dropTrailingAssistantPrefill(modifiedBody.messages);
+    }
+
     // Config-driven strip of params unsupported by the target provider/model.
     // For GitHub Copilot this removes Claude-style `thinking` and
     // `reasoning_effort` for Claude models that reject them upstream
@@ -188,6 +200,19 @@ export class GithubExecutor extends BaseExecutor {
     return { ...msg, content: cleanContent.length > 0 ? cleanContent : null };
   }
 
+  // Remove trailing assistant message(s). GitHub Copilot's /chat/completions endpoint
+  // can't honor an assistant prefill and 400s unless the conversation ends with a
+  // non-assistant (user/tool) message. Never empties the array — an assistant-only
+  // conversation keeps its last message. No-op (same array reference) when the
+  // conversation already ends with a non-assistant message.
+  // Port of 9router#2143 (author: Manuel <baslr@users.noreply.github.com>).
+  dropTrailingAssistantPrefill(messages: any): any {
+    if (!Array.isArray(messages) || messages.length === 0) return messages;
+    let end = messages.length;
+    while (end > 1 && messages[end - 1]?.role === "assistant") end--;
+    return end === messages.length ? messages : messages.slice(0, end);
+  }
+
   async execute(input: ExecuteInput) {
     const result = await super.execute(input);
     if (!result || !result.response) return result;
diff --git a/tests/unit/executor-github-prefill-sanitize.test.ts b/tests/unit/executor-github-prefill-sanitize.test.ts
new file mode 100644
index 0000000000..346882951b
--- /dev/null
+++ b/tests/unit/executor-github-prefill-sanitize.test.ts
@@ -0,0 +1,119 @@
+import test from "node:test";
+import assert from "node:assert/strict";
+
+import { GithubExecutor } from "../../open-sse/executors/github.ts";
+
+// GitHub Copilot's /chat/completions endpoint rejects a conversation that ends with an
+// assistant message: "This model does not support assistant message prefill. The
+// conversation must end with a user message." Anthropic clients (e.g. newest Claude
+// Desktop) send a trailing assistant turn as a prefill seed — the Anthropic API honors
+// it, but Copilot 400s. GithubExecutor.dropTrailingAssistantPrefill() strips the
+// trailing assistant message(s) before dispatch, scoped to the github executor only.
+// Port of 9router#2143 (author: Manuel <baslr@users.noreply.github.com>).
+
+test("dropTrailingAssistantPrefill drops a single trailing assistant message", () => {
+  const executor = new GithubExecutor();
+  const messages = [
+    { role: "user", content: "Hi" },
+    { role: "assistant", content: "Here is the answer:" },
+  ];
+
+  const out = executor.dropTrailingAssistantPrefill(messages);
+
+  assert.equal(out.length, 1);
+  assert.equal(out[0].role, "user");
+});
+
+test("dropTrailingAssistantPrefill drops multiple consecutive trailing assistant messages", () => {
+  const executor = new GithubExecutor();
+  const messages = [
+    { role: "user", content: "Hi" },
+    { role: "assistant", content: "one" },
+    { role: "assistant", content: "two" },
+  ];
+
+  const out = executor.dropTrailingAssistantPrefill(messages);
+
+  assert.equal(out.length, 1);
+  assert.equal(out[0].role, "user");
+});
+
+test("dropTrailingAssistantPrefill is a no-op (same reference) when the conversation ends with a user message", () => {
+  const executor = new GithubExecutor();
+  const messages = [
+    { role: "user", content: "Hi" },
+    { role: "assistant", content: "Hello" },
+    { role: "user", content: "More" },
+  ];
+
+  const out = executor.dropTrailingAssistantPrefill(messages);
+
+  assert.equal(out, messages, "must return the same array reference when nothing changes");
+  assert.equal(out.length, 3);
+});
+
+test("dropTrailingAssistantPrefill is a no-op when the conversation ends with a tool message", () => {
+  const executor = new GithubExecutor();
+  const messages = [
+    { role: "user", content: "Hi" },
+    { role: "assistant", content: null, tool_calls: [{ id: "x" }] },
+    { role: "tool", tool_call_id: "x", content: "result" },
+  ];
+
+  const out = executor.dropTrailingAssistantPrefill(messages);
+
+  assert.equal(out, messages, "must return the same array reference when nothing changes");
+  assert.equal(out.length, 3);
+  assert.equal(out[2].role, "tool");
+});
+
+test("dropTrailingAssistantPrefill never empties an assistant-only conversation", () => {
+  const executor = new GithubExecutor();
+  const messages = [{ role: "assistant", content: "only" }];
+
+  const out = executor.dropTrailingAssistantPrefill(messages);
+
+  assert.equal(out.length, 1, "must keep at least one message");
+  assert.equal(out[0].role, "assistant");
+});
+
+test("dropTrailingAssistantPrefill is null/empty safe", () => {
+  const executor = new GithubExecutor();
+
+  assert.deepEqual(executor.dropTrailingAssistantPrefill([]), []);
+  assert.equal(executor.dropTrailingAssistantPrefill(undefined), undefined);
+  assert.equal(executor.dropTrailingAssistantPrefill(null), null);
+});
+
+test("GithubExecutor.transformRequest drops the trailing assistant prefill end-to-end", () => {
+  const executor = new GithubExecutor();
+  const body = {
+    model: "claude-sonnet-4.6",
+    messages: [
+      { role: "user", content: "Hi" },
+      { role: "assistant", content: "Here is the answer:" },
+    ],
+  };
+
+  const out = executor.transformRequest("claude-sonnet-4.6", body, false, {});
+
+  assert.equal(out.messages.length, 1);
+  assert.equal(out.messages[0].role, "user");
+});
+
+test("GithubExecutor.transformRequest leaves a user-terminated conversation untouched end-to-end", () => {
+  const executor = new GithubExecutor();
+  const body = {
+    model: "claude-sonnet-4.6",
+    messages: [
+      { role: "user", content: "Hi" },
+      { role: "assistant", content: "Hello" },
+      { role: "user", content: "More" },
+    ],
+  };
+
+  const out = executor.transformRequest("claude-sonnet-4.6", body, false, {});
+
+  assert.equal(out.messages.length, 3);
+  assert.equal(out.messages[2].role, "user");
+});
diff --git a/tests/unit/executor-github.test.ts b/tests/unit/executor-github.test.ts
index a4ae4eb4d3..cf4efb0006 100644
--- a/tests/unit/executor-github.test.ts
+++ b/tests/unit/executor-github.test.ts
@@ -112,6 +112,11 @@ test("GithubExecutor.transformRequest injects JSON response instructions for Cla
         reasoning_text: "internal",
         reasoning_content: "internal",
       },
+      // Trailing user turn: dropTrailingAssistantPrefill (9router#2143) strips a
+      // conversation that ends in "assistant", which would otherwise remove the very
+      // message this test inspects below. Keep the array ending in "user" so this test
+      // stays focused on response_format injection + reasoning-field stripping.
+      { role: "user", content: "thanks" },
     ],
   };
 
@@ -225,6 +230,11 @@ test("GithubExecutor.transformRequest leaves string content and missing content
         role: "assistant",
         tool_calls: [{ id: "c1", type: "function", function: { name: "f", arguments: "{}" } }],
       },
+      // Trailing tool response: dropTrailingAssistantPrefill (9router#2143) strips a
+      // conversation that ends in "assistant", which would otherwise remove the very
+      // tool_calls message this test inspects below. A real tool round-trip ends in
+      // "tool", not "assistant" — model that shape instead.
+      { role: "tool", tool_call_id: "c1", content: "result" },
     ],
   };
   const result = executor.transformRequest("claude-sonnet-4.6", body, true, {});