Documentation
Insidia tests your AI app and the application around it in one local run. Start with a running app on localhost, then review the evidence in the HTML report. The CLI is free and pre-release.
Run your first scan
Section titled “Run your first scan” Run it yourself Install the CLI, configure your app, and run an L1 scan.
Use a coding agent Give your agent a local scope and have it set up the scan with you.
After the scan
Section titled “After the scan” Read your report Understand execution status, policy results, evidence, and reruns.
Understand coverage See what L1 checks and where the current release has limits.
Find a specific answer
Section titled “Find a specific answer”- Troubleshooting: resolve setup errors and incomplete scans.
- Run in CI: automate L1 and keep reports when a scan fails.
- Configuration: target types, request formats, credentials, and model endpoints.
- CLI reference: commands, flags, and exit codes.
- Scope: which hosts a scan may contact.
- Engines: install optional tools and choose coverage.
- Benchmark: how the scanner is evaluated.
- Data handling: what stays local and what is stored in a run.
Insidia Cloud, the hosted attacker, is planned for December 2026. It is not required to use the local CLI.