redact-secret · Report

API key (unprefixed)

32 alphanumeric characters recognised only beside a same-line AI21 name, host or SDK constructor; the shape rests on one ten-sample maintainer observation.

AI21 Labs
Detectors: ai21-api-key
Run 2026-10-07
Mode published · redact-secret 0.1.0-beta.14

All providers

Dossier verdictReady
Dossier evidence levelT2 · Tool-corroborated
Dossier researched2026-09-29

Registry detector ai21-api-key since redact-secret#868 (registry pinned at cfe2aec); graduated from a Beta.10 arrival family (#384, research #784), contract in benchmarks/lib/beta8/384e.ts. T0, context-gated: no provider statement and no scanner rule; pending.

New pipelineAuthority

Built from the new pipeline

The new pipeline is the authority for credential qualification. These numbers are read from the qualification view derived from the official credential-eval run of the public-evidence-snapshot population. The regression and policy populations keep their own counts on the qualification pages and are not added in here.

Authority: new
Population: public-evidence-snapshot
Engine: credential-eval 0.1.0-alpha.16
Evidence: snapshot-2026.10.06.4
Run: sha256:4bec6e539482…
Recorded: 2026-10-07
Authorised for: @redact-secret/core@0.1.0-beta.14

Maintainer-reviewed (independent review pending)

598 fixtures of the 7,041 in evidence release snapshot-2026.10.06.4 carry this label. They are counted in the numbers on this page.

Passed the project’s own verification (source evidence, automated checks, recorded counter-arguments); not yet independently reviewed.

OpenRedaction default (all patterns): not measured in this run (optional)

The default (all-pattern) profile is slow and noisy for credential evaluation: personal data and resource identifiers dominate its findings, and its last official measurement took 7 minutes of scanner time on a population the credential profile scans in seconds (its methods run takes 40+ minutes), so it is a manual measurement and never blocks other scanners or core verification (#763). Its earlier results stay as history with their run identity; the credential profile is a separate scanner, never spliced with it.

Last measurement: policy-corpus@linux-x64 (configuration sha256:2451ce7bf380), public-evidence-snapshot+methods@linux-x64 (configuration sha256:d47d081ce5fb), public-evidence-snapshot@linux-x64 (configuration sha256:2451ce7bf380), regression-corpus@linux-x64 (configuration sha256:2451ce7bf380) · engine 0.1.0-alpha.5 · recorded 2026-10-05 · retained record, no longer listed by the run registry. It stays labelled with that run identity and is never combined with another profile or another run.

Open the qualification page for this pointer

OpenRedaction credential profile (33 types): not measured in an official run (local exploratory diagnostics only: see ADR)

The credential profile is the comparison profile chosen in docs/decisions/2026-10-06-choose-the-openredaction-credential-bearing-profile-as-the-comparison-scanner.md. It is optional in the official run class because no pinned engine configuration measures it yet: making it required would refuse every official view until the owner approves the one profile-only run. Its own history starts with that run; nothing of the default's is reused for it.

No earlier measurement of it is recorded. Nothing is shown in its place.

Official measurement: pending: requires owner approval (one profile-only run; see the decision).

Open the qualification page for this pointer

Every population and its qualification

Research record

From credential-evidence snapshot-2026.10.06.4 · records at 77ce761 · schema 1.8.0. It describes the research on the format, not what any scanner or the product does, and not a support status.

Review stateDraft, not reviewed
Format revision1 · current
ResearchResearched
Researched2026-09-29

3 events in the review history: 2 observed, 1 reviewed. Latest: observed on 2026-09-29 by automation, project maintainer. Project-maintained review is not independent validation. The family record at this release.

Format

Provider format research from credential-evidence snapshot-2026.10.06.4 · records at 77ce761 · schema 1.8.0.

What it looks like

Descriptive pattern: ^[A-Za-z0-9]{32}$

Parts are shown as recorded. Evidence classes belong to the facts below; no class is assigned to a part.

Format facts

Tool corroborated · tool-corroboration · current · observed 2026-09-26
Pinned scanner rules are consistent with the contract grammar (3 artifacts: LLM-Runner-Router (independent implementation); Vulnetix vnx-sec-315; keychecker (independent implementation)).
- cunnymessiah/keychecker @ 122eadc6bdbbeec9f31e94878f98c4692d5b5718: main.pyscanner-rule-source · last read 2026-09-29 · latest outcome read · supports keychecker (independent implementation): [A-Za-z0-9]{32} filter before a live AI21 check (2023-12-16) · #L242
- MCERQUA/LLM-Runner-Router @ dc020cfeb6ae071d5582bed6ce3563eea64ae8fd: src/auth/BYOKManager.jsscanner-rule-source · last read 2026-09-29 · latest outcome read · supports LLM-Runner-Router (independent implementation): ai21 keyFormat '[A-Za-z0-9]{32}' (2025-08-26) · #L287-L290
- Vulnetix/cli @ c6e24fc9b0148dc0a227da70774300fba7f339ee: internal/sast/rules/vnx-sec-315.regoscanner-rule-source · last read 2026-09-29 · latest outcome read · supports Vulnetix vnx-sec-315: ai21[_-]?api[_-]?key assignment + exactly [A-Za-z0-9]{32} (2026-06-14) · #L25
Tool corroborated · field-shape · current · observed 2026-09-26
shape: 32 alphanumeric characters, no prefix (One peer rule and two independent implementations (redact-secret#1013); no provider statement. AI21-committed literals agree but are held for ruling Q-AI.)
- cunnymessiah/keychecker @ 122eadc6bdbbeec9f31e94878f98c4692d5b5718: main.pyscanner-rule-source · last read 2026-09-29 · latest outcome read · supports a 32-alphanumeric filter · #L242
- MCERQUA/LLM-Runner-Router @ dc020cfeb6ae071d5582bed6ce3563eea64ae8fd: src/auth/BYOKManager.jsscanner-rule-source · last read 2026-09-29 · latest outcome read · supports keyFormat '[A-Za-z0-9]{32}' · #L287-L290
- github.com/redact-secret/redact-secret/issues/784scanner-rule-source · last read 2026-09-26 · latest outcome read · supports ten-sample maintainer observation, values withheld
- redact-secret/redact-secret @ add1188fed9993723c59fbce8c867086b9d2049a: docs/audits/evidence/1013/ai21-api-key.mdscanner-rule-source · last read 2026-09-29 · latest outcome read · supports shape: 32 alphanumeric characters, no prefix
- Vulnetix/cli @ c6e24fc9b0148dc0a227da70774300fba7f339ee: internal/sast/rules/vnx-sec-315.regoscanner-rule-source · last read 2026-09-29 · latest outcome read · supports exactly [A-Za-z0-9]{32} after an ai21 key name · #L25
Provider documented · field-context · current · observed 2026-09-26
context: a same-line AI21_API_KEY name, api.ai21.com host or AI21 SDK constructor
- AI21Labs/ai21-python @ main: ai21/ai21_env_config.pyprovider-sdk-source · last read 2026-09-26 · latest outcome read · supports _ENV_API_KEY = "AI21_API_KEY"
- AI21Labs/ai21-python @ main: ai21/http_client/base_http_client.pyprovider-sdk-source · last read 2026-09-26 · latest outcome read · supports Authorization: Bearer {api_key}
Unresolved · field-prefix-and-checksum · current · observed 2026-09-26
prefix-and-checksum: no prefix or checksum was observed or documented
- github.com/redact-secret/redact-secret/issues/784scanner-rule-source · last read 2026-09-26 · latest outcome read · supports prefix-and-checksum: no prefix or checksum was observed or documented
- redact-secret/redact-secret @ add1188fed9993723c59fbce8c867086b9d2049a: docs/audits/evidence/1013/ai21-api-key.mdscanner-rule-source · last read 2026-09-29 · latest outcome read · supports prefix-and-checksum: no prefix or checksum was observed or documented
Unresolved · field-other-widths · current · observed 2026-09-26
other-widths: a UUID (Kingfisher, deleted 2026-08-21) or 40-64 alphanumerics (octocode) (Excluded by the exact-32 contract; no fixture asserts either way.)
- redact-secret/redact-secret @ add1188fed9993723c59fbce8c867086b9d2049a: docs/audits/evidence/1013/ai21-api-key.mdscanner-rule-source · last read 2026-09-29 · latest outcome read · supports both readings contradict every AI21 literal found
Provider documented · field-non-secrets · current · observed 2026-09-26
non-secrets: AI21_API_HOST, AI21_API_VERSION, AI21_AWS_REGION and the console's masked suffix are not credentials
- AI21Labs/ai21-python @ main: ai21/ai21_env_config.pyprovider-sdk-source · last read 2026-09-26 · latest outcome read · supports non-secrets: AI21_API_HOST, AI21_API_VERSION, AI21_AWS_REGION and the console's masked suffix are not credentials
Unresolved · listed-references · current · observed 2026-09-26
The legacy contract lists 7 references without stating which property each supports.
- docs.ai21.com/docs/create-api-keyprovider-documentation · last read 2026-09-29 · latest outcome read · supports Listed as a reference by the legacy contract
- docs.litellm.ai/docs/providers/ai21other · last read 2026-09-26 · latest outcome read · supports Listed as a reference by the legacy contract
- github.com/redact-secret/redact-secret/issues/868issue-or-discussion · last read 2026-09-26 · latest outcome read · supports Listed as a reference by the legacy contract
- github.com/redact-secret/redact-secret/issues/784scanner-rule-source · last read 2026-09-26 · latest outcome read · supports Listed as a reference by the legacy contract
- redact-secret/redact-secret @ add1188fed9993723c59fbce8c867086b9d2049a: docs/audits/evidence/1013/ai21-api-key.mdscanner-rule-source · last read 2026-09-29 · latest outcome read · supports Listed as a reference by the legacy contract
- AI21Labs/ai21-python @ main: ai21/ai21_env_config.pyprovider-sdk-source · last read 2026-09-26 · latest outcome read · supports Listed as a reference by the legacy contract
- AI21Labs/ai21-python @ main: ai21/http_client/base_http_client.pyprovider-sdk-source · last read 2026-09-26 · latest outcome read · supports Listed as a reference by the legacy contract
Tool corroborated · dossier-research · current · observed 2026-09-29
Legacy dossier research (verdict ready, tier T2) cited 5 sources; the dossier does not attribute sources to individual properties.
- AI21Labs/ai21-python @ bbc422f6f955134c2e33a9473ec0f71d21611764: examples/studio/batches/batches.pyother · last read 2026-09-29 · latest outcome read · supports Cited by the legacy dossier research for this family · #L6
- cunnymessiah/keychecker @ 122eadc6bdbbeec9f31e94878f98c4692d5b5718: main.pyscanner-rule-source · last read 2026-09-29 · latest outcome read · supports Cited by the legacy dossier research for this family · #L242
- MCERQUA/LLM-Runner-Router @ dc020cfeb6ae071d5582bed6ce3563eea64ae8fd: src/auth/BYOKManager.jsscanner-rule-source · last read 2026-09-29 · latest outcome read · supports Cited by the legacy dossier research for this family · #L287-L290
- redact-secret/redact-secret @ add1188fed9993723c59fbce8c867086b9d2049a: docs/audits/evidence/1013/ai21-api-key.mdscanner-rule-source · last read 2026-09-29 · latest outcome read · supports Final research evidence recorded by the legacy dossier
- Vulnetix/cli @ c6e24fc9b0148dc0a227da70774300fba7f339ee: internal/sast/rules/vnx-sec-315.regoscanner-rule-source · last read 2026-09-29 · latest outcome read · supports Cited by the legacy dossier research for this family · #L25
Unresolved · taxonomy-sources · current · observed 2026-09-29
The legacy taxonomy lists 1 source for this family. The taxonomy records no date; the dossier researchedAt is used as the observed-at date.
- docs.ai21.com/docs/create-api-keyprovider-documentation · last read 2026-09-29 · latest outcome read · supports Listed as a source for this family in the legacy taxonomy

Open questions

No open question is recorded for this revision.

Benchmark dossier notes

From the provider dossier, as written. The evidence level above says how well the format is backed; a fact the dossier does not record is not shown.

Shape: no prefix; exactly 32 characters [A-Za-z0-9], mixed case, recognised only beside a same-line AI21 name, host or SDK constructor (context-constrained, never bare). Not stated in provider prose.
Basis: T2 on the corroborated route (redact-secret#1013, 2026-09-29), context-constrained. Without any provider material, three references from three owners in two non-summary classes clear the bar: - Vulnetix vnx-sec-315 (peer-scanner-rule, 2026-06-14): ai21[_-]?api[_-]?key assignment + exactly 32 [A-Za-z0-9]; - cunnymessiah/keychecker (independent-implementation, 2023-12-16): a 32-alphanumeric filter before a live AI21 check; - MCERQUA/LLM-Runner-Router (independent-implementation, 2025-08-26): ai21: { keyFormat: '[A-Za-z0-9]{32}' }. AI21 staff also committed key literals to AI21Labs example code (ai21-python 2024-08 to 2025-03, one pointing at a staging host; a 2023 AI21 demo app's Bearer value): six literals, each 32 mixed alphanumerics, none a UUID. Their validity is unknown and was not tested. Whether a provider-committed literal of unknown validity counts as a provider example (R5) or as a leak is maintainer ruling Q-AI, pending; the T2 verdict holds without them, so they are recorded as bounded evidence pending Q-AI, not counted. Values are never reproduced. The 2026-09-27 ten-sample maintainer observation (#784) agrees and is kept as authoring input only.
Issuance: not attempted. The #784 checklist covers length, charset, fixed segments, case mix, the masked suffix shown afterwards, and a second workspace.
Contract in core: detector-families.md, section Keyword-gated provider keys (#868). #774 lists it as pending, needing "real issuance evidence, not more fixtures".

In this benchmark

Fixture rows on the current run. Counts are for redact-secret in published · redact-secret 0.1.0-beta.14 mode.

Fixtures: 60
Left readable: 0
Redacted too much: 0
False alarms: 0

60 fixtures: 43 must stay quiet, 17 record project policy. See every row

redact-secret fixture counts by evidence level
Evidence level	Fixtures	Left readable	Too much	False alarms
T2Tool-corroborated	18	0	0	0
T3Project policy	42	0	0	0

Every scanner on the same fixtures

In run order. Counts are what each scanner recorded on this family's fixtures, whichever rules it has; a scanner with no rule for the family has nothing to report on it.

Counts per scanner on this family's fixtures
Scanner	Fixtures	Left readable	Too much	False alarms
flare-redactRuntime library · 1.6.1 · Published npm package · secrets-only (pii, generic_assignment disabled) · JavaScript engineNo rule maps to it	60	16	1	0
gitleaksRepository scanner · 8.30.1 · Directory scan · default rulesNo rule maps to it	60	0	0	1
redact-secretProduct measured here · 0.1.0-beta.14 · Published npm package · default detectors1 detector mapped	60	0	0	0
trufflehogRepository scanner · 3.97.4 · Filesystem scan · verification disabledNo rule maps to it	60	17	0	0

Looks like it, but isn't

Values the dossier records as resembling this credential without being one.

Collisions: MD5 and other 32-character hashes, UUIDs without dashes, session ids, other vendors' 32-byte keys, and the non-secret AI21_API_HOST, AI21_API_VERSION and AI21_AWS_REGION variables.

Scanner rules for this family

Mapped by hand (reviewed 2026-09-30) from each scanner's pinned rule file, never from what a scanner found on the fixtures.

No peer rule maps to this family

None of the reviewed peer scanners has a rule that can match a credential of this family.

None mapped

Find

Evidence level

Scanners

60 of 60 rows

Fixtures in this family

60 rows, redact-secret's outcome on each. Rows that need a look come first (0), then the rest in corpus order. Choose "Every scanner" to see each scanner's outcome for the same rows.

Fixtures in API key (unprefixed)
Fixture	Kind and evidence	redact-secret
`ai21-api-key-actions-env-literal`ai21 · documented-format-literal	Project policyT3 · Project policy	Redacted
`ai21-api-key-compose-env`ai21 · documented-format-literal	Project policyT3 · Project policy	Redacted
`ai21-api-key-curl-header`ai21 · documented-format-literal	Project policyT3 · Project policy	Redacted
`ai21-api-key-docker-run-env`ai21 · documented-format-literal	Project policyT3 · Project policy	Redacted
`ai21-api-key-dotenv`ai21 · documented-format-literal	Project policyT3 · Project policy	Redacted
`ai21-api-key-dotenv-alt`ai21 · documented-format-literal	Project policyT3 · Project policy	Redacted
`ai21-api-key-export`ai21 · documented-format-literal	Project policyT3 · Project policy	Redacted
`ai21-api-key-json-config`ai21 · documented-format-literal	Project policyT3 · Project policy	Redacted
`ai21-api-key-key-shape-bare`ai21 · documented-format-literal	Project policyT3 · Project policy	Redacted
`ai21-api-key-key-shape-quoted`ai21 · documented-format-literal	Project policyT3 · Project policy	Redacted
`ai21-api-key-key-shape-unicode-crlf`ai21 · documented-format-literal	Project policyT3 · Project policy	Redacted
`ai21-api-key-langchain-kwarg`ai21 · documented-format-literal	Project policyT3 · Project policy	Redacted
`ai21-api-key-litellm-yaml`ai21 · documented-format-literal	Project policyT3 · Project policy	Redacted
`ai21-api-key-other-host-curl-twin`ai21 · missing-context-marker	Must not flagT3 · Project policy · twin	Quiet
`ai21-api-key-proxy-log`ai21 · documented-format-literal	Project policyT3 · Project policy	Redacted
`ai21-api-key-python-ctor`ai21 · documented-format-literal	Project policyT3 · Project policy	Redacted
`ai21-api-key-tool-call`ai21 · documented-format-literal	Project policyT3 · Project policy	Redacted
`ai21-api-key-ts-ctor`ai21 · documented-format-literal	Project policyT3 · Project policy	Redacted
`ai21-api-key-actions-secret-reference`ai21 · templated-reference	Must not flagT3 · Project policy	Quiet
`ai21-api-key-angle-key-placeholder`ai21 · documentation-placeholder	Must not flagT3 · Project policy	Quiet
`ai21-api-key-bare-in-prose-near-miss`ai21 · format-near-miss	Must not flagT2 · Tool-corroborated	Quiet
`ai21-api-key-bare-line-near-miss`ai21 · format-near-miss	Must not flagT2 · Tool-corroborated	Quiet
`ai21-api-key-base64-text-encoded-value`ai21 · benign-encoded-value	Must not flagT2 · Tool-corroborated	Quiet
`ai21-api-key-data-uri-encoded-value`ai21 · benign-encoded-value	Must not flagT2 · Tool-corroborated	Quiet
`ai21-api-key-docs-ctor-placeholder`ai21 · documentation-placeholder	Must not flagT3 · Project policy	Quiet
`ai21-api-key-embedded-run-near-miss`ai21 · format-near-miss	Must not flagT2 · Tool-corroborated	Quiet
`ai21-api-key-env-reference-reference`ai21 · templated-reference	Must not flagT3 · Project policy	Quiet
`ai21-api-key-id-named-compose-twin`ai21 · missing-context-marker	Must not flagT3 · Project policy · twin	Quiet
`ai21-api-key-id-named-env-alt-twin`ai21 · missing-context-marker	Must not flagT3 · Project policy · twin	Quiet
`ai21-api-key-id-named-env-twin`ai21 · missing-context-marker	Must not flagT3 · Project policy · twin	Quiet
`ai21-api-key-id-named-export-twin`ai21 · missing-context-marker	Must not flagT3 · Project policy · twin	Quiet
`ai21-api-key-id-named-json-twin`ai21 · missing-context-marker	Must not flagT3 · Project policy · twin	Quiet
`ai21-api-key-id-named-kwarg-twin`ai21 · missing-context-marker	Must not flagT3 · Project policy · twin	Quiet
`ai21-api-key-id-named-tool-call-twin`ai21 · missing-context-marker	Must not flagT3 · Project policy · twin	Quiet
`ai21-api-key-key-guidance-prose`ai21 · prose-mention	Must not flagT3 · Project policy	Quiet
`ai21-api-key-key-shape-bare-twin`ai21 · wrong-length	Must not flagT2 · Tool-corroborated · twin	Quiet
`ai21-api-key-key-shape-quoted-twin`ai21 · wrong-length	Must not flagT2 · Tool-corroborated · twin	Quiet
`ai21-api-key-key-shape-unicode-crlf-twin`ai21 · wrong-length	Must not flagT2 · Tool-corroborated · twin	Quiet
`ai21-api-key-label-prose`ai21 · benign-lookalike	Must not flagT3 · Project policy	Quiet
`ai21-api-key-long-value-twin`ai21 · wrong-length	Must not flagT2 · Tool-corroborated · twin	Quiet
`ai21-api-key-mask`ai21 · benign-lookalike	Must not flagT3 · Project policy	Quiet
`ai21-api-key-missing-keyword`ai21 · benign-lookalike	Must not flagT2 · Tool-corroborated	Quiet
`ai21-api-key-model-id-public-id`ai21 · public-identifier	Must not flagT2 · Tool-corroborated	Quiet
`ai21-api-key-model-notes-prose`ai21 · prose-mention	Must not flagT3 · Project policy	Quiet
`ai21-api-key-object-id-public-id`ai21 · public-identifier	Must not flagT2 · Tool-corroborated	Quiet
`ai21-api-key-other-host-log-twin`ai21 · missing-context-marker	Must not flagT3 · Project policy · twin	Quiet
`ai21-api-key-other-provider-model-twin`ai21 · missing-context-marker	Must not flagT3 · Project policy · twin	Quiet
`ai21-api-key-project-uuid-public-id`ai21 · public-identifier	Must not flagT2 · Tool-corroborated	Quiet
`ai21-api-key-python-environ-reference`ai21 · templated-reference	Must not flagT3 · Project policy	Quiet
`ai21-api-key-reference`ai21 · benign-lookalike	Must not flagT3 · Project policy	Quiet

Sources

Researched 2026-09-29.

Documentation and code

docs.ai21.com/docs/create-api-key
github.com/Vulnetix/cli/blob/c6e24fc9b0148dc0a227da70774300fba7f339ee/internal/sast/rules/vnx-sec-315.rego#L25
github.com/cunnymessiah/keychecker/blob/122eadc6bdbbeec9f31e94878f98c4692d5b5718/main.py#L242
github.com/MCERQUA/LLM-Runner-Router/blob/dc020cfeb6ae071d5582bed6ce3563eea64ae8fd/src/auth/BYOKManager.js#L287-L290
github.com/AI21Labs/ai21-python/blob/bbc422f6f955134c2e33a9473ec0f71d21611764/examples/studio/batches/batches.py#L6

Research log

redact-secret/redact-secret#784Research issue
redact-secret/redact-secret#774Research issue
redact-secret/redact-secret#868Research issue
redact-secret/redact-secret#1013Research issue
redact-secret/redact-secret-benchmarks#384Research issue
Final evidence, pinned to a commit/redact-secret/redact-secret/blob/add1188fed9993723c59fbce8c867086b9d2049a/docs/audits/evidence/1013/ai21-api-key.md