fix(cloudflare): normalize null content to empty string for tool round-trips

Cloudflare's OpenAI-compat endpoint rejects assistant messages with
content: null, even when tool_calls are present (standard OpenAI format).
Added normalizeMessages() that converts null content to "" before dispatch,
plus a regression test covering the null-content + tool_calls case.

Also credits @moaaz12-web in README Contributors for the tool-calling PR.

README.md

@@ -245,11 +245,17 @@ Contributors very welcome! Good first PRs:
 ```bash
 npm install
 npm run dev      # server on :3001, dashboard on :5173, both with HMR
-npm test         # vitest — 69 tests across providers, routes, router, ratelimit
+npm test         # vitest — 75 tests across providers, routes, router, ratelimit
 ```
 
 PRs should include a test, keep the existing test suite green, and match the `.editorconfig` / tsconfig defaults already in the repo. Issues and discussions are open.
 
+### Contributors
+
+Thanks to everyone who's helped improve FreeLLMAPI:
+
+- [@moaaz12-web](https://github.com/moaaz12-web) — tool-calling support across providers (#3)
+
 ## Terms of Service review
 
 A self-hosted, single-user, personal-use setup was reviewed against each provider's ToS (April 2026). Summary:

server/src/__tests__/providers/cloudflare.test.ts

@@ -53,4 +53,43 @@ describe('CloudflareProvider', () => {
       provider.chatCompletion('no-colon-here', [{ role: 'user', content: 'Hi' }], 'model')
     ).rejects.toThrow(/account_id:api_token/);
   });
+
+  it('should convert null assistant content to empty string (CF rejects null)', async () => {
+    let capturedBody: any = null;
+    vi.spyOn(global, 'fetch').mockImplementation(async (_url, init) => {
+      capturedBody = JSON.parse((init as any).body);
+      return {
+        ok: true,
+        json: () => Promise.resolve({
+          id: 'chatcmpl-cf',
+          object: 'chat.completion',
+          created: 123,
+          model: '@cf/meta/llama-3.1-70b-instruct',
+          choices: [{ index: 0, message: { role: 'assistant', content: 'ok' }, finish_reason: 'stop' }],
+          usage: { prompt_tokens: 1, completion_tokens: 1, total_tokens: 2 },
+        }),
+      } as any;
+    });
+
+    await provider.chatCompletion(
+      'abc123:token',
+      [
+        { role: 'user', content: 'Weather?' },
+        {
+          role: 'assistant',
+          content: null,
+          tool_calls: [{
+            id: 'call_1',
+            type: 'function',
+            function: { name: 'get_weather', arguments: '{"city":"Karachi"}' },
+          }],
+        },
+        { role: 'tool', tool_call_id: 'call_1', content: '{"temp":30}' },
+      ],
+      '@cf/meta/llama-3.1-70b-instruct',
+    );
+
+    expect(capturedBody.messages[1].content).toBe('');
+    expect(capturedBody.messages[1].tool_calls).toHaveLength(1);
+  });
 });

server/src/providers/cloudflare.ts

@@ -20,6 +20,14 @@ export class CloudflareProvider extends BaseProvider {
     return { accountId: apiKey.slice(0, sep), token: apiKey.slice(sep + 1) };
   }
 
+  // Cloudflare's OpenAI-compat endpoint rejects `content: null` on assistant
+  // messages that carry tool_calls, even though the OpenAI spec allows it.
+  private normalizeMessages(messages: ChatMessage[]): ChatMessage[] {
+    return messages.map(m =>
+      m.content === null ? { ...m, content: '' } : m,
+    );
+  }
+
   async chatCompletion(
     apiKey: string,
     messages: ChatMessage[],
@@ -37,7 +45,7 @@ export class CloudflareProvider extends BaseProvider {
       },
       body: JSON.stringify({
         model: modelId,
-        messages,
+        messages: this.normalizeMessages(messages),
         temperature: options?.temperature,
         max_tokens: options?.max_tokens,
         top_p: options?.top_p,
@@ -74,7 +82,7 @@ export class CloudflareProvider extends BaseProvider {
       },
       body: JSON.stringify({
         model: modelId,
-        messages,
+        messages: this.normalizeMessages(messages),
         temperature: options?.temperature,
         max_tokens: options?.max_tokens,
         top_p: options?.top_p,