Client error '400 Bad Request' for url '<LLM_BASE_URL>/responses'

[hrivera-ntap]

Issue

I have a suspicion this is because the tool call descriptions are empty. Might be related to this: openai/codex#37692

Here's a simple way to repro:

<USER>@<HOST>:~/dummy/foo (main) % cecli --version
cecli 1.2.1

<USER>@<HOST>:~/dummy/foo (main) % cecli --agent --model openai/gpt-5.6-luna --message "create a hello.sh script that echos hello. Validate that script prints 'hello'"       
───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────


4.0k ◇ 77.1% ↑ 92 ↓ $0.00036 • 4.1k ◇ 77.1% ↑↓ $0.00036
───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Tool Call: Local • UpdateTodoList
Remaining:
→ Inspect repository and determine script location/content requirements
○ Create executable hello.sh that prints hello
○ Validate hello.sh output is exactly hello
Client error '400 Bad Request' for url '<LLM_BASE_URL>/responses'
For more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/400
Retrying in 0.2 seconds...


4.2k ◇ 96.0% ↑ 70 ↓ $0.00020 • 8.3k ◇ 86.7% ↑↓ $0.00055
───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Tool Call: Local • ReadFile

range_1: .gitignore • @000 • 000@

✓ Retrieved context for 1 operation(s)


4.5k ◇ 93.0% ↑ 49 ↓ $0.00020 • 13k ◇ 89.0% ↑↓ $0.00076
───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Tool Call: Local • ResourceManager
create: hello.sh

⛭ Modifying Context
🗀 Created 'hello.sh' and made it editable


4.6k ◇ 96.6% ↑ 68 ↓ $0.00020 • 18k ◇ 91.0% ↑↓ $0.00096
───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Tool Call: Local • EditFile

file_path:
hello.sh

replacement:
@@ -0,0 +1,2 @@
+#!/bin/sh
+printf '%s\n' 'hello'


4.8k ◇ 96.9% ↑ 85 ↓ $0.00022 • 22k ◇ 92.3% ↑↓ $0.0012
───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Tool Call: Local • UpdateTodoList
Done:
✓ Inspect repository and determine script location/content requirements
✓ Create executable hello.sh that prints hello

Remaining:
○ Validate hello.sh output is exactly hello
Client error '400 Bad Request' for url '<LLM_BASE_URL>/responses'
For more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/400
Retrying in 0.2 seconds...
Client error '400 Bad Request' for url '<LLM_BASE_URL>/responses'
For more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/400
Retrying in 0.5 seconds...
Client error '400 Bad Request' for url '<LLM_BASE_URL>/responses'
For more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/400
Retrying in 1.0 seconds...
Client error '400 Bad Request' for url '<LLM_BASE_URL>/responses'
For more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/400
Retrying in 2.0 seconds...


4.9k ◇ 96.7% ↑ 66 ↓ $0.00021 • 27k ◇ 93.1% ↑↓ $0.0014
───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Tool Call: Local • Command

Command:
chmod +x hello.sh && output=$(./hello.sh) && test "$output" = "hello" && printf 'output=%s\n' "$output" && test -x hello.sh && printf 'executable=yes\n'
Allow execution of this command? (Y)es/(N)o/(A)ll/(S)kip all/(D)on't ask again [Yes]:

⛭ Executing shell command with 30s timeout.


5.1k ◇ 97.5% ↑ 64 ↓ $0.00020 • 33k ◇ 93.8% ↑↓ $0.0016
───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Tool Call: Local • UpdateTodoList
Done:
✓ Inspect repository and determine script location/content requirements
✓ Create executable hello.sh that prints hello
✓ Validate hello.sh output is exactly hello
Client error '400 Bad Request' for url '<LLM_BASE_URL>/responses'
For more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/400
Retrying in 0.2 seconds...


5.2k ◇ 97.3% ↑ 68 ↓ $0.00021 • 38k ◇ 94.3% ↑↓ $0.0018
───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Tool Call: Local • Yield

Summary:
Created executable `hello.sh` containing a portable shell shebang and a `printf` command that prints exactly `hello` followed by a newline. Validation succeeded: `./hello.sh` output was exactly 
`hello`, and the file has executable permissions.

Commit 0521d6d feat: add hello script that prints hello
<USER>@<HOST>:~/dummy/foo (main) % echo $?
1

Version and model info

cecli v1.2.1
model: openai/gpt-5.6-luna