Skip to content

OpenAI's Codex Security CLI and TypeScript SDK for finding, validating, and fixing security vulnerabilities. npm: https://www.npmjs.com/package/@openai/codex-security

2.7 KB
Raw Blame

Import ArmorCode findings

Import a saved response from ArmorCode's POST /api/findings endpoint into the local workbench. Cloud publication of ArmorCode findings is unverified. The reader accepts the API's data.findings envelope, as well as a top-level findings array for exported reports. Each finding needs a stable numeric or string id and a title. Reuse the same source key for repeat imports from the same ArmorCode organization and finding selection.

Choose SAST findings for one repository in ArmorCode. Set ARMORCODE_QUERY to the path of the JSON request body for that selection. Severity and status filters alone can return findings from multiple repositories. From the checkout root, save the response in a private temporary directory outside the checkout.

report_dir=$(mktemp -d "${TMPDIR:-/tmp}/armorcode.XXXXXX")
curl --fail-with-body -X POST "https://$HOST_URL/api/findings" \
  -H 'accept: application/json' \
  -H "authorization: Bearer $ARMORCODE_API_KEY" \
  -H 'content-type: application/json' \
  --data-binary "@$ARMORCODE_QUERY" > "$report_dir/armorcode-findings.json"

Preview the same file against the selected checkout.

codex-security publish findings "$report_dir/armorcode-findings.json" \
  --to local --repository . --provider armorcode \
  --source-key ORG_ID/APP_ID --dry-run --format json

If any SAST row omits its repository URL, verify the report's repository in ArmorCode and add --source-repository-url URL to the command. The import checks declared repository URLs against the checkout's configured remotes. A conflicting repository in the report causes the import to fail.

The import preserves the complete source record as evidence. It maps severity, description, mitigation, finding URLs, CVE identifiers, file path and line number, and the last-updated timestamp when present. The evidence URL uses the source scanner's url when present. ArmorCode's findingUrl remains in the retained source record. Findings whose scanType contains SAST also expose their source status, scanner origin and verdict, rule IDs, CWE IDs, and mitigation as typed SAST details. A record's sourceRepositoryUrl, repositoryUrl, or repository URL is projected when present. Unknown severities are retained as unknown. The local assessment starts as not_assessed. Repeated IDs in a single input are excluded as ambiguous. Repeat imports with the same provider, source key and finding ID update the existing observation.

Review the preview's findings, exclusions, repository and source key. Remove --dry-run to save the findings, then run codex-security findings imported . --format json to read them back. Records omitted from a later import remain stored. Use the normalized field names above when an export uses different field names.