When running Shannon via `npx @keygraph/shannon start`, the CLI consistently times out with "Timeout waiting for workflow to start" before the worker container can register with Temporal and begin processing.

[premkumar-sudo]

Describe the bug

When running Shannon via npx @keygraph/shannon start, the CLI consistently times out with "Timeout waiting for workflow to start" before the worker container can register with Temporal and begin processing.
Environment

• OS: Ubuntu 24.04 LTS (GCP VM, e2-standard-8)
• Docker: 27.x
• Node.js: v20.20.2
• Shannon: v1.0.0 (via npx)
• Provider: Google Vertex AI (us-east5)

Steps to reproduce

1.npx @keygraph/shannon setup (configure Vertex AI)
2. npx @keygraph/shannon start -u <url> -r <repo> -c <config>
3. CLI shows banner, then "Waiting for workflow to start..." with dots
4. After ~60s, "Timeout waiting for workflow to start"

Expected behaviour

Worker should start and workflow should begin processing.

Actual behaviour

The CLI times out, but running the worker container manually with docker run using the same config works perfectly — preflight passes, pre-recon agent launches, and Claude starts analyzing.

Pre-submission checklist (required)

• I have searched the existing open issues and confirmed this bug has not already been reported.
• I am running the latest released version of shannon.

If applicable

• I have included relevant error messages, stack traces, or failure details.
• I have checked the workspaces folder for logs and pasted the relevant errors.
• I have inspected the failed Temporal workflow run and included the failure reason.
• I have included clear steps to reproduce the issue.
• I have redacted any sensitive information (tokens, URLs, repo names).

Debugging details

Running the worker container manually works:

docker run --rm --network shannon-net \
  -e TEMPORAL_ADDRESS=shannon-temporal:7233 \
  -e CLAUDE_CODE_USE_VERTEX=1 \
  ... \
  keygraph/shannon:1.0.0 \
  node apps/worker/dist/temporal/worker.js   \
  --task-queue shannon-pentest --config 

Screenshots

No response

CLI mode

npx (@keygraph/shannon)

Provider

Google Vertex AI

Full command with all flags used (with redactions)

npx @keygraph/shannon start -u https://example.com/ -r ~/repo-of-my-app -c ~/congif-file.yaml

OS (with version)

Ubuntu 24.04 LTS (GCP VM, e2-standard-8)

Node.js version ('node -v')

v20.20.2

Docker version ('docker -v')

27.x

Additional context

• The worker container spawned by the CLI (spawnWorker in index.mjs) uses --rm and stdio: "pipe", making it impossible to capture logs after timeout.
• The timeout seems hardcoded and may be too short for the Vertex AI initialization + workflow bundling (~15-20s).
• Tested on both a manually cloned repo (./shannon start) and npx — same behavior.
• Works fine on macOS with OrbStack.

[ezl-keygraph]

Thanks for reporting. Can you provide the following details which will be helpful for us to reproduce the issue on our side

1. Are you able to successfully run Shannon with Vertex AI on macOS?
2. Are you running npx @keygraph/shannon start as root or as a non-root user on the Linux VM?

[premkumar-sudo]

Are you able to successfully run Shannon with Vertex AI on macOS?
Yes

Are you running npx @keygraph/shannon start as root or as a non-root user on the Linux VM?
For both I have tried, and the result is same.

[ganipadela]

I am also facing similar issue on my macOS

[s4mpler]

Got the same behavior testing on Debian 13(ARM) and MacOS. Neither npx nor running directly from the Repo worked. The shannon-worker doesn't seem to start correctly or at all... also no logs to be found on what the could cause the error

[s4mpler]

A small update: I tried it again on a x86 platform in a LXC Container with fresh installed NodeJS 25.09 and I got it running. Seems to be related to Arm

[sudhir-s-in]

i tried it on x86 platform , with node -v as 25 but even it failed to start the workflow .
tested it both on npx and locally