Team X-Ray could not start the Copilot SDK and did not have a GitHub token available for the GitHub Models fallback. Your options:
- Install and authenticate the Copilot CLI:
curl -fsSL https://gh.io/copilot-install | bash copilot auth login - If the CLI is installed outside your PATH, set an absolute
teamxray.cliPathin user or remote settings. - Run
Team X-Ray: Set GitHub Tokenfrom the Command Palette.
The extension can't find the Copilot CLI. Make sure it's installed and on your PATH:
# Install
curl -fsSL https://gh.io/copilot-install | bash
# Authenticate
copilot auth login
# Verify
copilot --versionIf copilot --version works in a different shell but not from VS Code, set teamxray.cliPath to the absolute executable path in user or remote settings and reload VS Code.
In the current implementation, BYOK settings are applied through the Copilot SDK session. That means you still need a working Copilot CLI even when teamxray.aiProvider is set to byok-openai, byok-anthropic, or byok-azure.
Also verify:
- You ran
Team X-Ray: Set BYOK API Key (Secure) teamxray.byokBaseUrlis setteamxray.byokModelis set and valid for the selected provider (required for BYOK sessions)
The API key isn't being sent or isn't valid for the provider you selected.
- Re-run
Team X-Ray: Set BYOK API Key (Secure)and paste the key again. The secret is stored per-machine; it won't roam with your settings. - Confirm the key matches the provider: an
sk-ant-...key withteamxray.aiProviderset tobyok-openaiwill always 401. - If you're migrating from the deprecated
teamxray.byokApiKeysettings entry, remove it fromsettings.jsonafter setting the secure key. Team X-Ray reads BYOK keys only from SecretStorage.
Almost always a byokBaseUrl mistake.
- OpenAI: the URL needs the
/v1suffix:https://api.openai.com/v1, nothttps://api.openai.com. - Anthropic: same pattern:
https://api.anthropic.com/v1. - Azure OpenAI: the URL must include both the resource name and the deployment name. Get the full URL from the Azure Portal: OpenAI resource → Deployments → your deployment → "Target URI". It looks like
https://YOUR-RESOURCE.openai.azure.com/openai/deployments/YOUR-DEPLOYMENT.
If you're behind a corporate proxy or gateway, confirm the gateway path matches what the provider's SDK expects, not just the hostname.
The selected model ID is unavailable to your token or no longer appears in the catalog.
- Open the GitHub Models catalog and choose an available model ID.
- Set
teamxray.githubModelsModelto that ID. - Run the analysis again.
teamxray.aiProvider = github-models runs the direct API path. With copilot or a byok-* provider, GitHub Models is the fallback path.
Azure routes by deployment name, not model name. If you set teamxray.byokModel to gpt-4o, Azure will reject it unless your deployment happens to be named gpt-4o.
Set teamxray.byokModel to your deployment name from the Azure Portal Deployments list.
Large repos (300K+ commits) take time even with the worker thread. The worker prevents VS Code from freezing, but parsing hundreds of thousands of commits is inherently slow.
What to expect:
- Repos under 10K commits: seconds
- 10K–100K commits: under a minute
- 100K+ commits: a few minutes
If you're contributing and see module resolution errors with the Copilot SDK, check that dynamic imports use the webpack ignore directive:
const sdk = await import(/* webpackIgnore: true */ '@github/copilot-sdk');Without this, webpack tries to bundle the ESM module at build time and fails.
Check your VS Code version:
Help → About → Version
Team X-Ray requires VS Code 1.100.0 or later. Update if you're on an older version.
detectBot() matches contributor names and emails against known patterns:
- Names containing "bot", "dependabot", "renovate", "github-actions"
- Email domains associated with automated services
If a human contributor triggers bot detection, it's likely because their git config name or email matches one of these patterns. Update their git identity:
git config user.name "Actual Name"
git config user.email "real@email.com"By design. The webview adapts to your VS Code theme (light or dark). The exported HTML always uses the standalone X-Ray dark theme (#0a0a0f background, #06b6d4 cyan accent, scan-line effects) for a consistent look when sharing.
Team X-Ray commands won't appear in the Command Palette until the extension activates. Try:
- Open a folder that contains a git repository
- Reload VS Code: Command Palette →
Developer: Reload Window - Check the Extensions sidebar — make sure Team X-Ray is enabled