feat(enrichment): detect AI-provider and more SaaS/CI credential formats in secret-scan

review-enrichment/src/analyzers/secret-scan.ts

@@ -271,6 +271,102 @@ const RULES: Rule[] = [
     re: /\bfo1_[A-Za-z0-9_-]{43}\b/,
     confidence: "high",
   },
+  {
+    // Dropbox short-lived access token: `sl.` + 130-152 base64url chars.
+    kind: "dropbox_token",
+    re: /\bsl\.[A-Za-z0-9_-]{130,152}\b/,
+    confidence: "high",
+  },
+  {
+    // JFrog Artifactory API key: `AKCp8` + >=69 base62 (lookahead terminator — the body has no fixed end).
+    kind: "jfrog_api_key",
+    re: /\bAKCp8[A-Za-z0-9]{69,}(?![A-Za-z0-9])/,
+    confidence: "high",
+  },
+  {
+    // Duffel API token: `duffel_test_`/`duffel_live_` + 43 base64url.
+    kind: "duffel_token",
+    re: /\bduffel_(?:test|live)_[A-Za-z0-9_-]{43}\b/,
+    confidence: "high",
+  },
+  {
+    // EasyPost API key: `EZAK` (production) / `EZTK` (test) + 54 base62.
+    kind: "easypost_key",
+    re: /\bEZ[AT]K[A-Za-z0-9]{54}\b/,
+    confidence: "high",
+  },
+  {
+    // Frame.io developer token: `fio-u-` + 64 base64url.
+    kind: "frameio_token",
+    re: /\bfio-u-[A-Za-z0-9_-]{64}\b/,
+    confidence: "high",
+  },
+  {
+    // Contentful personal access token: `CFPAT-` + 43 base64url.
+    kind: "contentful_token",
+    re: /\bCFPAT-[A-Za-z0-9_-]{43}\b/,
+    confidence: "high",
+  },
+  {
+    // SonarQube token: `sqa_`/`sqp_`/`squ_` (analysis/project/user) + 40 hex.
+    kind: "sonarqube_token",
+    re: /\bsq[apu]_[a-f0-9]{40}\b/,
+    confidence: "high",
+  },
+  {
+    // Pulumi access token: `pul-` + 40 hex.
+    kind: "pulumi_token",
+    re: /\bpul-[a-f0-9]{40}\b/,
+    confidence: "high",
+  },
+  {
+    // Adafruit IO key: `aio_` + 28 base62.
+    kind: "adafruit_io_key",
+    re: /\baio_[A-Za-z0-9]{28}\b/,
+    confidence: "high",
+  },
+  {
+    // ReadMe API key: `rdme_` + >=70 lowercase-hex-ish body (lookahead terminator).
+    kind: "readme_api_key",
+    re: /\brdme_[a-z0-9]{70,}(?![a-z0-9])/,
+    confidence: "high",
+  },
+  {
+    // Typeform personal access token: `tfp_` + >=40 base62/._- (lookahead terminator).
+    kind: "typeform_token",
+    re: /\btfp_[A-Za-z0-9._-]{40,}(?![A-Za-z0-9._-])/,
+    confidence: "high",
+  },
+  {
+    // Sentry DSN: an ingest URL embedding a 32-hex public key against a *.sentry.io host + project id.
+    kind: "sentry_dsn",
+    re: /\bhttps:\/\/[a-f0-9]{32}@[a-z0-9.-]*sentry\.io\/[0-9]+\b/,
+    confidence: "high",
+  },
+  {
+    // New Relic license/ingest key: 40 hex with the `NRAL` suffix (distinct from the `NRAK-` user key above).
+    kind: "newrelic_license_key",
+    re: /\b[a-f0-9]{40}NRAL\b/,
+    confidence: "high",
+  },
+  {
+    // Replicate API token: `r8_` + 37 base62.
+    kind: "replicate_token",
+    re: /\br8_[A-Za-z0-9]{37}\b/,
+    confidence: "high",
+  },
+  {
+    // Groq API key: `gsk_` + 52 base62.
+    kind: "groq_api_key",
+    re: /\bgsk_[A-Za-z0-9]{52}\b/,
+    confidence: "high",
+  },
+  {
+    // Perplexity API key: `pplx-` + >=40 base62 (lookahead terminator).
+    kind: "perplexity_api_key",
+    re: /\bpplx-[A-Za-z0-9]{40,}(?![A-Za-z0-9])/,
+    confidence: "high",
+  },
   {
     kind: "private_key",
     re: /-----BEGIN (?:RSA |EC |OPENSSH |DSA |PGP )?PRIVATE KEY-----/,

review-enrichment/test/secret-scan.test.ts

@@ -442,3 +442,49 @@ test("scanPatch does not flag near-miss variants of the new SaaS/cloud credentia
     assert.equal(findings.length, 0, `near-miss should not match: ${nm}`);
   }
 });
+
+test("scanPatch flags additional high-confidence AI-provider and SaaS/CI credential formats", () => {
+  const cases = [
+    ["dropbox_token", "sl." + b62(140)],
+    ["jfrog_api_key", "AKCp8" + b62(70)],
+    ["duffel_token", "duffel_test_" + b62(43)],
+    ["easypost_key", "EZAK" + b62(54)],
+    ["frameio_token", "fio-u-" + b62(64)],
+    ["contentful_token", "CFPAT-" + b62(43)],
+    ["sonarqube_token", "sqp_" + hex(40)],
+    ["pulumi_token", "pul-" + hex(40)],
+    ["adafruit_io_key", "aio_" + b62(28)],
+    ["readme_api_key", "rdme_" + hex(70)],
+    ["typeform_token", "tfp_" + b62(40)],
+    ["sentry_dsn", "https://" + hex(32) + "@o0.ingest.sentry.io/12345"],
+    ["newrelic_license_key", hex(40) + "NRAL"],
+    ["replicate_token", "r8_" + b62(37)],
+    ["groq_api_key", "gsk_" + b62(52)],
+    ["perplexity_api_key", "pplx-" + b62(40)],
+  ];
+  for (const [kind, secret] of cases) {
+    const findings = scanPatch("src/config.ts", hunk([`const c = "${secret}";`]));
+    assert.equal(findings.length, 1, `${kind}: expected exactly one finding, got ${JSON.stringify(findings)}`);
+    assert.equal(findings[0].kind, kind, `${kind}: wrong kind`);
+    assert.equal(findings[0].confidence, "high", `${kind}: wrong confidence`);
+  }
+});
+
+test("scanPatch does not flag near-miss variants of the new AI/SaaS credential formats", () => {
+  // One char short of the fixed/minimum length must produce no finding.
+  const nearMisses = [
+    "AKCp8" + b62(68),
+    "duffel_test_" + b62(42),
+    "EZAK" + b62(53),
+    "CFPAT-" + b62(42),
+    "pul-" + hex(39),
+    "aio_" + b62(27),
+    "r8_" + b62(36),
+    "gsk_" + b62(51),
+    "sqp_" + hex(39),
+  ];
+  for (const nm of nearMisses) {
+    const findings = scanPatch("src/config.ts", hunk([`const c = "${nm}";`]));
+    assert.equal(findings.length, 0, `near-miss should not match: ${nm}`);
+  }
+});