Here's a concise and complete GitHub issue you can paste:
Describe the bug
Strix fails to connect to the OpenAI API despite a valid API key and a working Docker environment.
The application immediately exits with:
LLM CONNECTION FAILED
Could not establish connection to the language model.
Error: Connection error.
The error occurs before any scan begins.
To Reproduce
- Install Strix v1.5.0 on macOS (Apple Silicon).
- Install Docker CLI with Homebrew and run Docker using Colima.
- Set the environment variables:
export LLM_API_KEY="<valid OpenAI API key>"
export OPENAI_API_KEY="$LLM_API_KEY"
export STRIX_LLM="openai/gpt-5"
- Verify Docker works:
- Verify the OpenAI API works:
curl https://api.openai.com/v1/models \
-H "Authorization: Bearer $LLM_API_KEY"
- Run:
strix -n --target https://mitarbeiter.reederei-luedicke.de/onboard/
Expected behavior
Strix should successfully connect to the OpenAI API and begin the scan.
Actual behavior
Strix exits immediately with:
LLM CONNECTION FAILED
Could not establish connection to the language model.
Error: Connection error.
No additional logs are produced, even when running with:
STRIX_LOG_LEVEL=debug
DEBUG=1
or
strix -n --target ... 2>&1 | tee strix.log
System Information
- OS: macOS 15 (Apple Silicon, arm64)
- Runtime: Colima (Docker runtime)
- Docker: Docker Engine Community 29.7.2 (CLI), Server 29.5.2
- Strix Version: 1.5.0
- Python Version: Standalone binary (not using a local Python installation)
- LLM Used:
openai/gpt-5
Additional context
I verified the following before opening this issue:
docker run hello-world succeeds.
docker info succeeds.
- Colima is running and healthy.
curl https://api.openai.com/v1/models succeeds using the same API key.
- The OpenAI account has access to
gpt-5.
LLM_API_KEY and OPENAI_API_KEY are both set.
STRIX_LLM=openai/gpt-5.
- There is no
~/.strix/cli-config.json.
- Setting
OPENAI_API_BASE=https://api.openai.com/v1 does not change the behavior.
strix --update reports that v1.5.0 is already the latest version.
This appears to be a connection issue inside Strix rather than a Docker or OpenAI API configuration issue.
Here's a concise and complete GitHub issue you can paste:
Describe the bug
Strix fails to connect to the OpenAI API despite a valid API key and a working Docker environment.
The application immediately exits with:
The error occurs before any scan begins.
To Reproduce
curl https://api.openai.com/v1/models \ -H "Authorization: Bearer $LLM_API_KEY"Expected behavior
Strix should successfully connect to the OpenAI API and begin the scan.
Actual behavior
Strix exits immediately with:
No additional logs are produced, even when running with:
or
System Information
openai/gpt-5Additional context
I verified the following before opening this issue:
docker run hello-worldsucceeds.docker infosucceeds.curl https://api.openai.com/v1/modelssucceeds using the same API key.gpt-5.LLM_API_KEYandOPENAI_API_KEYare both set.STRIX_LLM=openai/gpt-5.~/.strix/cli-config.json.OPENAI_API_BASE=https://api.openai.com/v1does not change the behavior.strix --updatereports that v1.5.0 is already the latest version.This appears to be a connection issue inside Strix rather than a Docker or OpenAI API configuration issue.