General
What does StretchDev actually do? It measures the quality of your code. You connect repositories as projects, run scans against their source, and get A–E ratings (Maintainability, Reliability, Security), a duplication percentage, a full issue list, and a pass/fail quality gate — plus a developer API platform for keys, webhooks, and logs.
Does StretchDev run my code? No. The analysis engine reads your source as text and measures it. It never executes your code, so scanning is always safe.
Is my data isolated from other organizations? Yes. Projects, scans, issues, API keys, and webhooks are all scoped to your workspace's organization automatically. You never pass an organization ID yourself, and you only see your own data.
Scans, ratings, and gates
Why does a new scan say "engine is not enabled"? Running a new scan is a compute action that requires the analysis engine. If it's off, you'll see a message like "StretchDev engine is not enabled. Set STRETCHDEV_ENGINE=python." Ask an administrator to enable it. Reading existing scans, ratings, and issues works regardless.
Why did my scan fail with "Provide files[] to scan"? No source was supplied. A scan needs at least one file path and its text.
What do the A–E ratings mean? A is best, E is worst, on three separate axes: Maintainability (code smells and duplication), Reliability (bugs), and Security (vulnerabilities). A project can be strong on one axis and weak on another — read all three.
How do I raise a rating? Fix the issue type behind it: close vulnerabilities to raise Security, fix bugs to raise Reliability, clear code smells and cut duplication to raise Maintainability. Then re-scan.
Why did my scan fail its quality gate? At least one gate condition wasn't met. Open the scan to see which condition failed (for example, Security below A or duplication over the threshold). Fix the underlying issue and re-scan, or adjust the policy if it's stricter than your team's real standard.
Can I test a gate change without running a scan? Yes. Preview the gate against the project's latest stored scan on the Quality tab — it recomputes the result without writing a new scan. Note the preview needs the engine enabled.
"This project has no scans to evaluate yet." A gate preview needs at least one stored scan for its metrics. Run a scan first.
Release risk and AI review
The Release-risk panel says intelligence is unavailable. Risk scoring recomputes and needs the analysis engine; ask an administrator to enable it.
How do I lower a release-risk score? Shrink the change (less churn), avoid touching hotspots, and add tests — raising the tests-changed count is the cheapest lever.
The AI reviewer says it's disabled or not ready. The AI code-review feature is optional and flag-gated. If it's off or the model is unavailable, StretchDev tells you plainly and the rest of the app keeps working. Ask an administrator to enable it.
I submitted code for AI review and got nothing back. If the code field was empty, there's nothing to review. If the model was momentarily unavailable, try again shortly.
API platform
I lost my API key secret. The full secret is shown only once at creation; StretchDev stores only a hash and a short prefix. Rotate the key to get a new secret, then update your integration.
My integration is suddenly getting rejected (401/403). The key may be paused, revoked, expired, or missing a required scope. Check its status and scopes in the API console, or rotate it. Confirm the fix in Request logs.
My webhook isn't firing. Confirm the subscription is active (not paused/disabled) and subscribed to the right event, then send a Test delivery and check the Deliveries log for the response status.
A webhook delivery shows a non-2xx response. Your receiving endpoint rejected or errored on the payload. Inspect the delivery detail and your endpoint logs, fix the endpoint, and re-test.
What does the Usage view cover? A rolling 30-day window of your organization's active keys, active webhooks, deliveries, requests, average latency, errors, and breakdowns by status and scope.
Common error messages
- "Project not found." / "Scan not found." The ID or key doesn't match a record in your organization. Confirm you're in the right workspace and it wasn't deleted.
- "A project name is required." Name a project before saving.
- "An API key name is required." / "A webhook name is required." / "A target URL is required." Fill in the missing required field.
- "Webhook subscription not found." / "API key not found." / "Request log not found." The referenced record isn't in your organization.
Still stuck?
If a compute feature (scan, gate preview, release risk, or AI review) is unavailable, the fix is almost always enabling the analysis engine or the AI reviewer — an administrator task. Everything read-only (projects, scans, ratings, stored issues, logs, usage) works without them, so you can keep reviewing history while the engine is switched on.
Was this helpful?
Help us improve this article
Use these controls to share whether this answer solved the issue. Feedback helps prioritize updates to StretchSuite Support.