Source checks plus requested SCA/DAST stages, artifacts and one CI exit.
keryn ci [path] --out <dir>Options
Section titled “Options”Usage: keryn ci [path] --out <new-directory> [--base <ref-or-result>] [--rules <packs>] [--policy <file>] Source analysis runs by default with Deep; unavailable or failed Deep returns incomplete (3). --no-deep explicitly chooses pattern checks only. --dependencies <file> supply a build-produced resolved list to the source scan (repeatable, without an application profile). --ai-baseline <receipt> --ai-baseline-suite <suite> compare the current AI run with reproduced before evidence --capabilities <config> --capabilities-base <snapshot> --fail-on-expansion gate configuration authority changes --sca-collect collect all OSV packages/advisories in one offline invocation --sca-context <deployment> bind explicit AI roles; --sca-snapshot <file> and --sca-as-of <UTC ISO> preserve freshness context --dast-ownership <challenge.json> required with --dast-active; fresh proof before execution --dast-auth <auth.json> header-env authenticated ZAP discovery and verification --ai-tests <suite.json> run authenticated AI security regressions; suite bytes pinned at preflight --application <id> --revision <id> optional caller binding; must match the AI test suite when supplied --sca-inventory <report> import OSV JSON/CycloneDX inventory instead of a vulnerability report --sca-bindings <file> explicit AI roles for --sca-inventory; --sca-inventory-complete attests input report collection --sca also run installed OSV-Scanner offline (--sca-online permits advisory requests) --sca-report <report> import SCA instead; --sca-parser <id>; --sca-completion complete|partial|unknown|failed --dast-url <url> run installed ZAP spider/passive checks; --dast-active opts into active checks; --dast-ajax adds the AJAX spider --dast-report <report.xml> import instead; requires --dast-url and --dast-parser burp|zap --dast-completion complete|partial|unknown|failed import completion (default unknown) --application-file <file> run every stage an application profile (keryn.dev/application/v1) declares: each source (+ secrets) and its dependencies, each artifact through artifact analyze, each web target through dast scan. Default: <path>/.keryn/application.yml when present; --no-application-file ignores it. A profile replaces --sca*/--dast-url/--base; --revision labels the run; evidence.json binds it profile-declared. Packages (.deb/.rpm/AppImage/archives) are unpacked and every member analysed. A source whose manifests are locked by a workspace lockfile above it (Cargo, npm/pnpm/yarn/bun) is covered by one stage per such lockfile, scanned once and attributed; a Deno project gets a keryn sca deno stage; a source with no manifest or lockfile gets no dependency stage ("no dependency inputs"). Each source is scanned as part of its repository (scan --repo-root): rule paths:, roles, include/exclude and the repository's .keryn.yml see repository-relative paths; dependency roles too. sources[n].deepTimeout (seconds) gives one source's Deep longer than the 300 s default; a source whose Deep stops at the limit is incomplete and its error names that key. Active web targets require a locally minted ownership challenge and fresh verification before execution. --max-artifact-size <bytes> per-artifact cap for profile artifacts without maxBytes (default 64 MiB; see artifact --help) --jobs <n> stages at a time, each in a child process (default a quarter of the cores, at most 4; 1-64). Outputs are the same bytes at any --jobs. Progress (start, end, every 30 s) goes to stderr. --fail-on error|warning|info|none (default error, or the profile's output.failOn); --json prints the aggregate summary. Writes per-stage artifacts, summaries, ledger.json (the merged findings ledger, see keryn ledger --help), evidence.json digests (covering ledger.json; timing.json files are listed as undigested) and associations.json with explicit evidence bases. evidence.json and summary.json carry the CLI's build provenance; summary.md states a dirty or unrecorded build. Only requested analyzers run. --base records changes; the gate still evaluates current findings, not only additions. Fetch the base revision in CI. Exit precedence: 4 execution > 2 usage/config > 3 incomplete > 1 findings > 10 warnings > 0 pass.Examples
Section titled “Examples”# Source analysis with cross-file analysis; if cross-file analysis is incomplete, it exits 3.keryn ci . --out out/ci --base origin/main
# Add dependency and web stages.keryn ci . --out out/ci --sca --dast-url https://staging.example.com
# Run every stage an application profile declares.keryn ci . --out out/ci --application-file .keryn/application.yml --jobs 2Exit codes
Section titled “Exit codes”| Code | Meaning |
|---|---|
| 0 | pass |
| 1 | findings |
| 2 | usage or configuration error |
| 3 | incomplete |
| 4 | execution failure |
| 10 | warnings |