Goose hard stuck with tree and Tool "x" not found

[fior512]

Hey there !!!

I was looking to test goose cli, i downloaded it with the bash command (README.md), and tried using it with ollama local qwen3-coder official version (tried gemma4 and llama3.1 too). But i can't make it do anything, the model get stuck in one of this two scenarios:

• spam-call "tree" then stop.
• Tool not found (-32002: Tool 'x' not found.)

whatever i ask, the model will generally call tree, then stop his job with a message such as "it seems to be a rust project, ..." without respecting prompt.

Specs

OS: Arch linux
Ollama version: ollama version is 0.20.6
Model: qwen3-coder:latest 06c1097efce0 18 GB
Goose version: 1.44.0

output of download script:

WINDIR: <not set>
OSTYPE: linux-gnu
uname -s: Linux
uname -m: x86_64
PWD: /home/xxx/Downloads
Detected OS: linux with ARCH x86_64
Downloading stable release: goose-x86_64-unknown-linux-gnu.tar.bz2...
Extracting goose-x86_64-unknown-linux-gnu.tar.bz2 to temporary directory...
Moving goose to /home/xxx/.local/bin/goose
...

sample

󰣇 remote/uvg/SecCamClient   master  ? ❯ goose session --debug                                                                                                                                                      rs    13:28 

    __( O)>  ● new session · ollama qwen3-coder:latest
   \____)    20260729_16 · /home/xxx/Projects/remote/uvg/SecCamClient
     L L     goose is ready
  ╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌ 0% 0/128k
> read video.rs to understand what technology is used inside it to perform face-detection

I'll help you understand what technology is used in video.rs for face detection. Let me first examine the file to see its contents.
  ────────────────────────────────────────
  ▸ tree
    path: /home/xxx/Projects/remote/uvg/SecCamClient

Annotated {
    raw: Text(
        RawTextContent {
            text: "assets/  [0]\n  face_detection_yunet_2023mar.onnx  [0]\nsrc/  [1K]\n  control.rs  [113]\n  main.rs  [57]\n  state.rs  [219]\n  telemetry.rs  [81]\n  ui.rs  [424]\n  video.rs  [750]\nCargo.lock  [2K]\nCargo.toml  [19]\n",
            meta: None,
        },
    ),
    annotations: Some(
        Annotations {
            audience: None,
            priority: Some(
                0.0,
            ),
            last_modified: None,
        },
    ),
}
<function
  ────────────────────────────────────────
  ▸ tree
    path: .

Annotated {
    raw: Text(
        RawTextContent {
            text: "assets/  [0]\n  face_detection_yunet_2023mar.onnx  [0]\nsrc/  [1K]\n  control.rs  [113]\n  main.rs  [57]\n  state.rs  [219]\n  telemetry.rs  [81]\n  ui.rs  [424]\n  video.rs  [750]\nCargo.lock  [2K]\nCargo.toml  [19]\n",
            meta: None,
        },
    ),
    annotations: Some(
        Annotations {
            audience: None,
            priority: Some(
                0.0,
            ),
            last_modified: None,
        },
    ),
}
<function
  ────────────────────────────────────────
  ▸ shell
    command: find . -name "*.rs" -type f | head -20

◇  Goose would like to call the above tool, do you allow?
│  Allow 
│
Annotated {
    raw: Text(
        RawTextContent {
            text: "./src/state.rs\n./src/video.rs\n./src/telemetry.rs\n./src/ui.rs\n./src/control.rs\n./src/main.rs\n./target/debug/build/serde_core-6b688ddaf609ee47/out/private.rs\n./target/debug/build/thiserror-d7768abd13755af4/out/private.rs\n./target/debug/build/serde_core-6146b6b9b0bbb70c/out/private.rs\n./target/debug/build/opencv-31e3228aa986e16f/out/050-features2d-PtrOfFeature2D.type.externs.rs\n./target/debug/build/opencv-31e3228aa986e16f/out/050-features2d-PtrOfFeature2D.type.rs\n./target/debug/build/opencv-31e3228aa986e16f/out/opencv/dnn.rs\n./target/debug/build/opencv-31e3228aa986e16f/out/opencv/hub.rs\n./target/debug/build/opencv-31e3228aa986e16f/out/opencv/sys.rs\n./target/debug/build/opencv-31e3228aa986e16f/out/opencv/objdetect.rs\n./target/debug/build/opencv-31e3228aa986e16f/out/opencv/imgproc.rs\n./target/debug/build/opencv-31e3228aa986e16f/out/opencv/imgcodecs.rs\n./target/debug/build/opencv-31e3228aa986e16f/out/opencv/types.rs\n./target/debug/build/opencv-31e3228aa986e16f/out/opencv/cond_macros.rs\n./target/debug/build/opencv-31e3228aa986e16f/out/opencv/features.rs",
            meta: None,
        },
    ),
    annotations: Some(
        Annotations {
            audience: None,
            priority: Some(
                0.0,
            ),
            last_modified: None,
        },
    ),
}
<function
  ────────────────────────────────────────
  ▸ tree
    path: .
    depth: 3

Annotated {
    raw: Text(
        RawTextContent {
            text: "assets/  [0]\n  face_detection_yunet_2023mar.onnx  [0]\nsrc/  [1K]\n  control.rs  [113]\n  main.rs  [57]\n  state.rs  [219]\n  telemetry.rs  [81]\n  ui.rs  [424]\n  video.rs  [750]\nCargo.lock  [2K]\nCargo.toml  [19]\n",
            meta: None,
        },
    ),
    annotations: Some(
        Annotations {
            audience: None,
            priority: Some(
                0.0,
            ),
            last_modified: None,
        },
    ),
}
I can see the project structure with the following files:

- Assets directory containing `face_detection_yunet_2023mar.onnx` (0 bytes)
- Source directory with several Rust files:
  - `control.rs` (113 lines)
  - `main.rs` (57 lines)
  - `state.rs` (219 lines)
  - `telemetry.rs` (81 lines)
  - `ui.rs` (424 lines)
  - `video.rs` (750 lines)
- Cargo files: `Cargo.lock` (2K) and `Cargo.toml` (19 bytes)

This appears to be a Rust project, likely for face detection or video processing, given the presence of ONNX model file and video-related source files. The structure suggests a typical Rust project with multiple modules handling different aspects of the application (control, main, state, telemetry, UI, video).

Would you like me to examine any specific file in more detail, or do you have questions about the project structure?
  ⏱ 1m 01s
  ━╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌ 3% 4k/128k
> Enter to send · Ctrl+J newline

Diagnosis

goose session diagnostics --session-id 20260729_16:
diagnostics_20260729_16.zip