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

[luciferlive112116]

Summary

The secret-scan analyzer (review-enrichment/src/analyzers/secret-scan.ts) flags credentials committed in a
PR diff, citing file:line + kind only (never the value). This adds 14 more credential formats the scanner
currently misses — the current wave of AI-provider keys plus widely-used SaaS/CI tokens — continuing the
established feat(enrichment) secret-format additions.

Supersedes the closed #3172. That review correctly flagged that a base64url token body terminated with \b
silently misses a real token whose last character is - (a non-word char). This version terminates every
base64url-body rule (dropbox_token, duffel_token, frameio_token, contentful_token) with a
negative-lookahead (?![A-Za-z0-9_-]) instead — the same terminator the existing SendGrid/Anthropic/Square
rules use — and drops the two formats (newrelic_license_key, replicate_token) whose exact length I could
not verify with confidence, rather than ship an imprecise rule.

Each new rule is a high-confidence, gitleaks/trufflehog-standard shape with a distinctive multi-character
literal prefix (or fixed marker/host) and a fixed or tightly-bounded length + charset, so the false-positive
rate against ordinary source, base64 blobs, hex hashes, and UUIDs is effectively zero:

kind	shape
groq_api_key	gsk_ + 52 base62
replicate_token	r8_ + 37 base62
perplexity_api_key	pplx- + ≥40 base62
dropbox_token	sl. + 130–152 base64url
jfrog_api_key	AKCp8 + ≥69 base62
duffel_token	duffel_{test,live}_ + 43
easypost_key	EZAK/EZTK + 54 base62
frameio_token	fio-u- + 64 base64url
contentful_token	CFPAT- + 43 base64url
sonarqube_token	sqa_/sqp_/squ_ + 40 hex
pulumi_token	pul- + 40 hex
adafruit_io_key	aio_ + 28 base62
readme_api_key	rdme_ + ≥70
typeform_token	tfp_ + ≥40
sentry_dsn	ingest URL: 32-hex key @…sentry.io/<project id>

No existing rule is modified, so current findings are unchanged, and no analyzer descriptor changes (the
finding schema is unchanged) — so analyzer-metadata.json and the generated UI mirror are untouched.

No linked issue: additive detection-coverage for well-known credential formats, matching the established
feat(enrichment) secret-format additions; each rule is a self-evident real token shape with no public
API/schema/deploy surface change — fits the repo's preferred (not required) linked-issue policy.

Scope

• The PR title follows type(scope): short summary Conventional Commit format, for example fix(api): restore profile access checks.
• This PR is focused and does not mix unrelated backend, UI, MCP, docs, dependency, and deploy changes.
• This follows CONTRIBUTING.md and does not reintroduce GitHub Pages, VitePress, site/, or CNAME.
• I linked an issue, or this is small enough that the summary explains why an issue is not needed.

Validation

• git diff --check
• npm run typecheck
• npm run rees:test — the review-enrichment build + analyzer suite (see note below)
• npm run test:coverage (N/A — this file is in review-enrichment/, outside the root src/** Codecov scope)
• npm run ui:build
• npm audit --audit-level=moderate
• New or changed behavior has unit/integration tests for new branches, fallback paths, and sanitizer boundaries

If any required check was skipped, explain why:

• Ran locally: git diff --check (clean), the review-enrichment TypeScript build (exit 0), and the secret-scan
  analyzer test via node --test — 44/44 pass, including a table case that flags each of the 14 formats at
  high confidence (the four base64url-body fixtures deliberately END IN - so they fail on a \b terminator
  and pass only on the lookahead — a real regression guard for the feat(enrichment): detect AI-provider and more SaaS/CI credential formats in secret-scan #3172 defect), plus a near-miss case
  asserting one-character-short tokens produce no finding. All fixtures are assembled from string
  fragments so the test file never contains a contiguous secret literal. The change is confined to
  review-enrichment/, outside the src/** Codecov scope.
• Not run locally: the UI, root typecheck, and the metadata:check step of rees:test. This change only adds
  scan rules, not any analyzer descriptor, so the committed analyzer-metadata.json/UI mirror are unchanged (a
  local regeneration produces a zero-content diff) and metadata:check passes on CI (Linux). On this Windows
  dev box metadata:check reports a spurious line-ending difference; it fails identically on unmodified
  main. analyzer-metadata.json was NOT modified.

Safety

• No secrets, wallet details, hotkeys, coldkeys, user PATs, private keys, raw trust scores, private rankings, or private maintainer evidence are exposed.
• Public GitHub text stays sanitized, low-noise, and does not imply compensation guarantees or optimization tactics.
• Auth, cookie, CORS, GitHub App, Cloudflare, or session changes include negative-path tests.
• API/OpenAPI/MCP behavior is updated and tested where needed.
• UI changes use live API data or real empty/error/loading states, not production mock/demo fallbacks.
• Visible UI changes include a UI Evidence section below with JPG/JPEG or PNG screenshots arranged as organized, captioned, clickable thumbnails. SVG screenshots are not used as review evidence. Review-only screenshots or recordings are not committed to the repository.
• Public docs/changelogs are updated where needed; changelogs are only edited for release-prep PRs.

Notes

• Detection-only additions: 14 new high-confidence rules; no existing rule, threshold, or descriptor changed,
  so current findings and analyzer-metadata.json are unaffected. The scanner still returns only file:line +
  kind, never the matched secret value.
• Each regex uses \b boundaries for word-character-terminated bodies (hex/base62), and a negative-lookahead
  terminator (?![A-Za-z0-9_-]) for base64url bodies that can legitimately end in -. sentry_dsn is
  anchored by its sentry.io host so it cannot match a bare URL.

[superagent-security]

Superagent didn't find any vulnerabilities or security issues in this PR.

[loopover-orb]

Warning

🟨🟨🟨🟨🟨🟨🟨🟨🟨🟨🟨🟨

⏸️ Gittensory review result - manual review recommended

_{Review updated: 2026-07-04 19:08:14 UTC}

2 files · 1 AI reviewer · no blockers · readiness 80/100 · CI green · clean

⏸️ Suggested Action - Manual Review

Review summary
This change cleanly extends the secret-scan rule table with additional high-confidence credential shapes and covers the main positive path for each new kind through scanPatch. The base64url-ending regression from the prior version is handled at the rule layer with negative lookahead terminators, and the tests deliberately exercise tokens ending in '-' for the affected formats. I do not see a reachable correctness break in the provided diff, but the PR text and coverage around boundary cases could be tightened.

Nits — 5 non-blocking

• nit: review-enrichment/test/secret-scan.test.ts:531 only exercises the negative-lookahead terminator for some new variable-length rules; add explicit overrun/terminator near-misses for dropbox_token, frameio_token, contentful_token, readme_api_key, typeform_token, and perplexity_api_key so we prove the rule stops at the right boundary instead of just matching the happy path.
• nit: review-enrichment/src/analyzers/secret-scan.ts:320 says ReadMe is a lowercase-hex-ish body, but the regex allows every lowercase letter plus digits; either tighten the charset if the format is actually hex or update the comment so future reviewers do not treat this as a verified hex-only shape.
• nit: PR description still lists replicate_token in the table even though this diff intentionally does not add a replicate_token rule; clean that up so the description matches the code under review.
• review-enrichment/test/secret-scan.test.ts:549 add near-misses like `"pplx-" + b62(39)` and `"rdme_" + hex(69)` to cover the minimum-length branches added in review-enrichment/src/analyzers/secret-scan.ts:325 and :341.
• review-enrichment/test/secret-scan.test.ts:531 add one valid `duffel_live_` case, since the regex accepts both `test` and `live` markers and the current test only drives the `test` branch.

Signal	Result	Evidence
Code review	✅ No blockers	1 reviewer
Linked issue	✅ Linked	#3172
Related work	✅ No active overlap found	No same-issue or scoped active PR overlap found.
Change scope	✅ 20/20	Low review scope from cached public metadata (1 linked issue).
Validation posture	❌ 5/25	Preflight is holding this PR: the review lane is unavailable, so it is not ready for automated review.
Contributor workload	✅ 10/10	Author activity: 119 registered-repo PR(s), 60 merged, 23 issue(s).
Contributor context	✅ Confirmed Gittensor contributor	luciferlive112116; Gittensor profile; 119 PR(s), 23 issue(s).
Gate result	✅ Passing	No configured blocker found.

Review context

• Author: luciferlive112116
• Role context: outside_contributor
• Public audience mode: oss maintainer
• Lane context: Repository registration is not available in the local Gittensory cache.
• Public profile languages: JavaScript, MDX, Python, Rust, TypeScript
• Official Gittensor activity: 119 PR(s), 23 issue(s).
• PR-specific overlap: none found.

Contributor next steps

• Await review-lane availability.
• Refresh registry data or choose a registered active repo.

Signal definitions

• Related work = same linked issue, overlapping active PRs, or title/path similarity.
• Change scope = cached public metadata such as size labels, draft state, and review-burden hints.
• Validation posture = whether the PR provides enough public validation/test evidence for maintainer review.
• Contributor workload = public contributor activity and cleanup pressure, not a repo-wide quality failure.
• Contributor context = public GitHub/Gittensor identity context; non-Gittensor status is not a blocker.

_{🟩 Safe / merged · 🟦 Advisory · 🟨 Held for review · 🟥 Blocked / closed}

---

💰 Earn for open-source contributions like this. Gittensor lets GitHub contributors earn for the work they already do — register to start earning →.

Checked by Gittensory, a quiet PR intelligence layer for OSS maintainers.

• Re-run Gittensory review

[loopover-orb]

Gittensory approves — the gate is satisfied and CI is green.