Disable fast path — always run full agent loop

README.md

@@ -9,7 +9,6 @@ This is an Open WebUI **Pipe** that exposes Claude Code as a selectable model. E
 - **Full Claude Code agent loop** — Read, Write, Edit, Bash, Glob, Grep, WebSearch, WebFetch (configurable allowlist)
 - **Per-chat workspaces** — each `chat_id` gets a sandboxed working directory that persists across turns
 - **Dual auth** — bring your own Anthropic **API key** (pay-per-token) *or* a **Claude Pro/Max OAuth token** (bills against your subscription)
-- **Fast path for trivial messages** — when no tools are needed, bypasses the SDK and hits the Messages API directly (~300 ms instead of multi-second CLI cold start)
 - **Streaming UI** — tool calls render inline with previews; generated images/PDFs/CSVs surface as artifacts in the chat
 - **Configurable valves** — model, permission mode, tool allowlist, max turns, workspace root
 

claude_agent_pipe.py

@@ -1364,18 +1364,7 @@ class Pipe:
             yield "_No user message to send to Claude Code._"
             return
 
-        # Gate: fast path for simple Q&A, agent path for tasks needing
+        # Fast path disabled — always run the full agent loop.
-        # tools/files. Same model, different invocation cost. The user can
-        # always force with `/agent` or `/fast` at the start of their message.
-        if not _needs_agent(prompt, __files__):
-            async for chunk in self._run_fast(
-                body, __user__, __metadata__, __files__, __event_emitter__
-            ):
-                yield chunk
-            return
-
-        # Agent path: strip the `/agent` prefix (if any) so Claude doesn't
-        # see it as content.
         prompt = _strip_mode_prefix(prompt)
 
         chat_id = __chat_id__ or "default"