@blamejs/pki 0.8.11 → 0.8.12

package/CHANGELOG.md

@@ -4,6 +4,20 @@ All notable changes to `@blamejs/pki` are documented here. The format
 follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/); this
 project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## v0.8.12 — 2026-09-23
+
+A certification path whose policies map to each other validates instead of exhausting the policy-node budget.
+
+### Changed
+
+- The `pki.path.validate` result carries `validPolicyGraph` where it carried `validPolicyTree`, and the structure it holds is a graph rather than a tree. It is `{ nodes }`, a flat list linked by index, each entry `{ id, depth, validPolicy, qualifierSet, expectedPolicySet, parents, children }`, where `parents` and `children` hold the ids of other entries and the root is id 0. A caller that walked `.children` from a root and read `.parent` reads the list and follows ids instead. The list is flat because a graph handed back as a root to walk re-expands into the exponential structure the moment anything recurses into every child, `JSON.stringify` included.
+- `C.LIMITS.PATH_MAX_POLICY_NODES` is unchanged at 4096 and still fails closed with `path/policy-tree-cap`, and `opts.maxPolicyNodes` still tightens it. What it bounds has changed with the structure: the graph grows with the number of distinct policies and mappings the path actually carries, so the cap is reached by a path that names that many policies, not by a path that names a few and is several certificates long. A single certificate carrying 4096 distinct policies still reaches it.
+
+### Fixed
+
+- `pki.path.validate` accepts a conforming certification path whose certificate policies map to each other across several levels. Such a path was refused with `path/policy-tree-cap` once its policy tree exceeded `maxPolicyNodes`, which the tree reached at six certificates carrying four mutually mapped policies. Certificate policies are now processed as the RFC 9618 `valid_policy_graph`, whose size is linear in the policies and mappings on the path rather than exponential in its length.
+- `userConstrainedPolicySet` reports the policies of the nodes whose only parent is the `anyPolicy` node, which is what RFC 5280 section 6.1.5 (g)(iii)(1) and RFC 9618 section 5.5 step (g)(2) both define it as. It reported the policies of the deepest nodes instead. The two answers differ wherever a policy mapping is in play: for a chain where an issuer domain policy is mapped to a subject domain policy, the set now names the issuer domain policy, which is the domain the relying party asked about and the reason mapping exists.
+
 ## v0.8.11 — 2026-09-23
 
 What a certificate, CRL, request or attribute certificate asserts in its extensions is read in one call.

package/README.md

@@ -265,7 +265,7 @@ comment blocks, is at [pkijs.com](https://pkijs.com).
 | `pki.cmc` | Build and interpret CMC messages (RFC 5272, obsoleted by RFC 10002). `build(spec, signer)` assembles a Full PKI Request across all three request arms (PKCS#10, CRMF, other) and carries it either way RFC 5272 §3.2 allows: `{ cert, key }` signs it, while `{ mac: { identifier, secret } }` authenticates it under a shared secret for a client that holds no key yet, using the PasswordRecipientInfo that section requires and deriving the key from the identifier followed by the secret. A key that cannot sign for itself proves possession another way: `spec.popChallenge` answers the authority's encrypted challenge (RFC 5272 §6.7), decrypting the proof value, confirming it hashes to the witness the challenge carries, and returning the MAC the authority asked for, with `cmc/pop-failed` when the value is not the one issued. A signed request goes through `pki.cms.sign` under `id-cct-PKIData`. A PKCS#10 (`tcr`) request is verified against its own proof-of-possession before signing, so a request whose self-signature does not verify under its subject public key is refused rather than signed into a message a CA would reject. Body-part identifiers are unique across the whole message and never the reserved 0; a caller's clash is refused rather than renumbered, since a control may already reference it. An Identity Proof V2 witness is computed over the `reqSequence` bytes exactly as emitted (§6.2.1 step 1) rather than a re-serialization, a POP Link Witness is emitted only alongside the POP Link Random control §6.3.1.1 requires beside it, and a renewal carries neither Identification nor Identity Proof in either version. `verify(response, sent)` takes what the CA returned plus the state the client retained and reduces it to one terminal outcome: `issued`, `pending`, `confirm-required`, `pop-required`, or `rejected`. It binds the exchange first — Transaction Identifier, the Sender and Recipient Nonce echo compared in constant time and by full value so a truncation cannot match, and the Data Return echo — with each check applying only if the client sent that half, and, once sent, an absent or differing echo being a refusal, which is the replay defense. `bodyPartIDs` extends the same rule to what the response is about: a status reporting on a body part the request never sent is refused, which the transaction and nonce cannot catch, because a server can echo both correctly while answering about a different message. Several status controls are permitted and the worst governs, so a failure cannot hide behind an earlier success; the absence of any status control is success, per §6.1.2. The carrier's signature must verify (§3.2.1.3.4): a conforming SignedData carries its own signer certificate, so the ordinary build-then-verify flow needs nothing extra and the verdict reports `signatureVerified: true`. Where the signer is found nowhere the posture is fail-closed with a named opt-out — supply `certs` with the responder's certificate, or `allowUnverified: true`, in which case the verdict reports `signatureVerified: false`. Doing neither is refused, the opt-out never excuses a signature that is present and wrong, and a carrier with no signer at all is refused outright. The response's own `cmsSequence` and `otherMsgs` come back raw, since a request whose only arm was the other-message form has no certificate to return and §4.1 puts its answer there. Nothing is trusted: issued certificates are read from the CMS certificate bag (§4.2) and surfaced raw for `pki.path.validate`, and a Publish Trust Anchors control is surfaced with `trusted: false` rather than added to a store — `build`, `verify` |
 | `pki.schema.cmc` | Decode CMC messages (RFC 5272 as updated by RFC 6402, the pair obsoleted by RFC 10002 / 10003 / 10004): a Full PKI Request (`PKIData`) or Full PKI Response (`PKIResponse`) riding inside a CMS SignedData, reached by the encapsulated content type (`id-cct-PKIData` / `id-cct-PKIResponse`). Controls are surfaced in wire order with their values raw, so an unrecognized one is data rather than a fault. Tagged requests decode across all three arms. The status verdicts (`CMCStatusInfo` v1 and `CMCStatusInfoV2`) are collected as an ordered list, and the RFC 6402 two-module `OtherStatusInfo` ambiguity — `pendInfo` and `extendedFailInfo` are both untagged SEQUENCEs in the 1988 module, told apart only by their first element — is resolved by inspection and refused when it cannot be told apart. Body-part identity is unique across the whole message rather than per sequence, 0 is reserved as the reference to the enclosing PKIData, and the `reqSequence` bytes are surfaced exactly as they appeared so an Identity Proof witness is computed over the wire bytes. A companion decoder for CMS content rather than an auto-routed format — `parse`, `parsePkiData`, `parsePkiResponse` |
 | `pki.schema.engine` | The declarative ASN.1 structure-schema engine every format parser composes — `walk`, `encode`, `embeddedDer`, and the schema combinators |
-| `pki.path` | Certification-path validation (RFC 5280 §6). `validate` runs the §6.1 state machine over an ordered path and a trust anchor: signature chaining across RSA, ECDSA, EdDSA, ML-DSA, SLH-DSA and hybrid composite ML-DSA (a composite is accepted only when both its post-quantum and traditional components verify), validity windows, name chaining, basic constraints and path length, key usage, name constraints, and the certificate-policy tree. It returns a structured verdict with per-check reason codes and enforces a `pki.trust` anchor's per-purpose distrust-after dates and delegator purposes through `checkPurpose` — an anchor that carries either one with no purpose named to select by is refused as a configuration fault rather than validated as though it carried none. The verdict says what was established. `revocationChecked` takes the weakest outcome on the path: `false` with no checker supplied, `"determined"` when every certificate got an explicit good or revoked answer, `"waived"` when `softFail` turned an undetermined one into a pass, and `"undetermined"` when one could not be answered at all — which includes a checker that throws, a fault `softFail` does not waive because `softFail` opts into an undetermined answer, not into a broken checker. `anchorConstraints` names the purpose the anchor was judged under and whether each of its two constraints applied. `crlChecker` supplies CRL-based revocation, covering partitioned and sharded CRLs, whose §6.3.3 Distribution Point ↔ IDP correspondence lets corresponding shards accumulate reason coverage until all eight revocation reasons are covered, and delta CRLs, merged onto the complete CRL they may be combined with (§5.2.4) so a held certificate its delta releases reads good, while a delta that merges with nothing still reports what it lists and still withholds good. `ocspChecker` supplies OCSP-based revocation (RFC 6960: CertID binding, responder authorization, signature, currency) over the same pluggable hook. `build(leaf, opts)` is the discovering complement (RFC 4158): from a leaf, an untrusted pool of candidate CA certificates, and a trust store, it finds the ordered leaf-to-anchor path `validate` accepts, using name chaining plus the RFC 4158 §3.5 sort hints (an AKI/SKI match, an anchor-adjacent issuer, CA plus keyCertSign, validity at the check time — ordering hints, never filters), a depth-first search with backtracking so the first accepted path wins, and a bounded search (chain-length cap, candidate-expansion cap, identity-tuple visited set) so a cross-certificate cycle or Bridge-CA fan-out terminates deterministically. Every accept flows through `validate`, and its verdict is cross-checked against `openssl verify`. Opt-in AIA `caIssuers` fetching (`opts.fetchAia: true`) discovers a missing intermediate from a certificate's Authority Information Access URL (§4.2.2.1) over `pki.transport`, triggered only on a pool miss and bounded against SSRF and amplification: HTTPS only, a total fetch budget that caps fetching silently rather than throwing, a per-cert URL cap, a build-wide URL dedupe, a response-size and certificate-count cap, no redirect following, and every fault a silent skip. The TLS trust (`opts.tls`) stays distinct from the PKI `trustAnchors`, and every fetched certificate remains untrusted pool material that still flows through `validate` (never a trust anchor). It is off by default, so the default build is byte-identical offline. An `opts.direction` of `"reverse"` searches from a trust anchor toward the leaf (RFC 4158 §3.1), narrower when the anchor set is small and the leaf's issuer options fan out, and `"auto"` picks the direction by first-hop fan-out; reverse building reuses the same pool, visited set, caps, and `validate` gate, so the direction changes only search order, and the anchor is excluded from the returned path. Reverse building is pool-only and does not combine with `fetchAia`. When `opts.userInitialPolicySet` names acceptable certificate-policy OIDs and more than one path validates, build returns the accepted path whose user-constrained policy set best satisfies that set (RFC 4158 §4) rather than the first accepted path, the ranking bounded by the same candidate cap and a malformed set refused with `path/bad-input`; an `anyPolicy` entry is unconstrained and returns the first accepted path. In a mesh of cross-certified domains, where many candidates chain by name to the same place, a candidate is ranked by how many leading relative distinguished names it shares with the name the search is heading for (§3.5.16 and the §3.5.19 reverse method), so a certificate in the target's own domain is tried before one from a cross-certified domain, and forward building sorts a candidate under every candidate that can complete the path when its path length constraint cannot cover the certificates already below it (§3.5.7), an ordering that outranks the other hints rather than competing with them. No candidate is removed and an unbounded search still reaches every path the pool holds; a bounded one spends `maxCandidatesConsidered` on whichever branches the order steers it to. `fetchingChecker(opts)` is the checker that goes and gets what it needs: it reads the cRLDistributionPoints and id-ad-ocsp locations out of the certificate and asks for them over `opts.transport`, OCSP first and a CRL after (RFC 5019 §3.1). It is off unless an operator builds it and passes it, so the default validation touches no network, and it runs during `validate` and not during `build`, because §3.2 has a status check requested only after the chain signatures are validated. A distribution point is fetched only as a `fullName` carrying a URI whose scheme this client speaks; ldap, ftp, a `nameRelativeToCRLIssuer` and a point naming a cRLIssuer and no location are counted as skips, so a certificate naming nothing fetchable reads as that rather than as a failed fetch. Nothing is trusted by having been fetched: a CRL goes to `crlChecker` and a response to `verifyOcspResponse`, and the structure a body parses as is decided by which extension named the URL, so a Content-Type mismatch is reported and never selects a parser. Every fault is an undetermined status carrying its reason and `check` throws nothing, which is what leaves `softFail` able to waive one. Plaintext `http` is accepted by default, these being signed objects the toolkit verifies, with `allowPlaintextHttp: false` to narrow it; the SSRF guards hold either way (no private, loopback or link-local destination, no redirect following, a response-size cap, a per-certificate URL cap, a total fetch count, and a wall-clock deadline across the validation). A verified response is cached in process on its CertID and responder URL and a CRL on its request URL and issuer name, with reuse decided from the signed thisUpdate and nextUpdate alone. Pure and re-entrant — `validate`, `build`, `crlChecker`, `ocspChecker`, `fetchingChecker` |
+| `pki.path` | Certification-path validation (RFC 5280 §6). `validate` runs the §6.1 state machine over an ordered path and a trust anchor: signature chaining across RSA, ECDSA, EdDSA, ML-DSA, SLH-DSA and hybrid composite ML-DSA (a composite is accepted only when both its post-quantum and traditional components verify), validity windows, name chaining, basic constraints and path length, key usage, name constraints, and the certificate policies, processed as the RFC 9618 `valid_policy_graph` in place of the RFC 5280 §6.1 policy tree. The graph holds one node per depth and policy identifier and gives it several parents where the tree made a separate node under each, so its size is linear in the policies and mappings on the path rather than exponential in the path length: a six-certificate chain whose policies map to each other needs 25 nodes where the tree needed 5461, and chains the tree could not hold inside `maxPolicyNodes` now validate. It returns a structured verdict with per-check reason codes and enforces a `pki.trust` anchor's per-purpose distrust-after dates and delegator purposes through `checkPurpose` — an anchor that carries either one with no purpose named to select by is refused as a configuration fault rather than validated as though it carried none. The verdict says what was established. `revocationChecked` takes the weakest outcome on the path: `false` with no checker supplied, `"determined"` when every certificate got an explicit good or revoked answer, `"waived"` when `softFail` turned an undetermined one into a pass, and `"undetermined"` when one could not be answered at all — which includes a checker that throws, a fault `softFail` does not waive because `softFail` opts into an undetermined answer, not into a broken checker. `anchorConstraints` names the purpose the anchor was judged under and whether each of its two constraints applied. `crlChecker` supplies CRL-based revocation, covering partitioned and sharded CRLs, whose §6.3.3 Distribution Point ↔ IDP correspondence lets corresponding shards accumulate reason coverage until all eight revocation reasons are covered, and delta CRLs, merged onto the complete CRL they may be combined with (§5.2.4) so a held certificate its delta releases reads good, while a delta that merges with nothing still reports what it lists and still withholds good. `ocspChecker` supplies OCSP-based revocation (RFC 6960: CertID binding, responder authorization, signature, currency) over the same pluggable hook. `build(leaf, opts)` is the discovering complement (RFC 4158): from a leaf, an untrusted pool of candidate CA certificates, and a trust store, it finds the ordered leaf-to-anchor path `validate` accepts, using name chaining plus the RFC 4158 §3.5 sort hints (an AKI/SKI match, an anchor-adjacent issuer, CA plus keyCertSign, validity at the check time — ordering hints, never filters), a depth-first search with backtracking so the first accepted path wins, and a bounded search (chain-length cap, candidate-expansion cap, identity-tuple visited set) so a cross-certificate cycle or Bridge-CA fan-out terminates deterministically. Every accept flows through `validate`, and its verdict is cross-checked against `openssl verify`. Opt-in AIA `caIssuers` fetching (`opts.fetchAia: true`) discovers a missing intermediate from a certificate's Authority Information Access URL (§4.2.2.1) over `pki.transport`, triggered only on a pool miss and bounded against SSRF and amplification: HTTPS only, a total fetch budget that caps fetching silently rather than throwing, a per-cert URL cap, a build-wide URL dedupe, a response-size and certificate-count cap, no redirect following, and every fault a silent skip. The TLS trust (`opts.tls`) stays distinct from the PKI `trustAnchors`, and every fetched certificate remains untrusted pool material that still flows through `validate` (never a trust anchor). It is off by default, so the default build is byte-identical offline. An `opts.direction` of `"reverse"` searches from a trust anchor toward the leaf (RFC 4158 §3.1), narrower when the anchor set is small and the leaf's issuer options fan out, and `"auto"` picks the direction by first-hop fan-out; reverse building reuses the same pool, visited set, caps, and `validate` gate, so the direction changes only search order, and the anchor is excluded from the returned path. Reverse building is pool-only and does not combine with `fetchAia`. When `opts.userInitialPolicySet` names acceptable certificate-policy OIDs and more than one path validates, build returns the accepted path whose user-constrained policy set best satisfies that set (RFC 4158 §4) rather than the first accepted path, the ranking bounded by the same candidate cap and a malformed set refused with `path/bad-input`; an `anyPolicy` entry is unconstrained and returns the first accepted path. In a mesh of cross-certified domains, where many candidates chain by name to the same place, a candidate is ranked by how many leading relative distinguished names it shares with the name the search is heading for (§3.5.16 and the §3.5.19 reverse method), so a certificate in the target's own domain is tried before one from a cross-certified domain, and forward building sorts a candidate under every candidate that can complete the path when its path length constraint cannot cover the certificates already below it (§3.5.7), an ordering that outranks the other hints rather than competing with them. No candidate is removed and an unbounded search still reaches every path the pool holds; a bounded one spends `maxCandidatesConsidered` on whichever branches the order steers it to. `fetchingChecker(opts)` is the checker that goes and gets what it needs: it reads the cRLDistributionPoints and id-ad-ocsp locations out of the certificate and asks for them over `opts.transport`, OCSP first and a CRL after (RFC 5019 §3.1). It is off unless an operator builds it and passes it, so the default validation touches no network, and it runs during `validate` and not during `build`, because §3.2 has a status check requested only after the chain signatures are validated. A distribution point is fetched only as a `fullName` carrying a URI whose scheme this client speaks; ldap, ftp, a `nameRelativeToCRLIssuer` and a point naming a cRLIssuer and no location are counted as skips, so a certificate naming nothing fetchable reads as that rather than as a failed fetch. Nothing is trusted by having been fetched: a CRL goes to `crlChecker` and a response to `verifyOcspResponse`, and the structure a body parses as is decided by which extension named the URL, so a Content-Type mismatch is reported and never selects a parser. Every fault is an undetermined status carrying its reason and `check` throws nothing, which is what leaves `softFail` able to waive one. Plaintext `http` is accepted by default, these being signed objects the toolkit verifies, with `allowPlaintextHttp: false` to narrow it; the SSRF guards hold either way (no private, loopback or link-local destination, no redirect following, a response-size cap, a per-certificate URL cap, a total fetch count, and a wall-clock deadline across the validation). A verified response is cached in process on its CertID and responder URL and a CRL on its request URL and issuer name, with reuse decided from the signed thisUpdate and nextUpdate alone. Pure and re-entrant — `validate`, `build`, `crlChecker`, `ocspChecker`, `fetchingChecker` |
 | `pki.identity` | Service identity (RFC 9525, which obsoletes RFC 6125). `match(cert, references, opts)` says whether a certificate presents a name the client was trying to reach, which is a separate answer from whether it chains: §1.2 has an application need both, and `pki.path.validate` takes `opts.identity` to run them together and report each. A reference identity is a domain name (DNS-ID), a textual IP address or a 4- or 16-octet `BufferSource` (IP-ID), `{ type: "srv", service: "_imap", value: "example.com" }` (SRV-ID, RFC 4985), or `{ type: "uri", scheme: "https", value: "www.example.com" }` (URI-ID). A record carries its service type and its domain together, so no comparison can pair the service of one reference with the domain of another (§6.5). The subject `commonName` is never read, under any option or fallback, including when the certificate carries no subjectAltName at all (§2). Names are compared label by label under an ASCII-only fold with one trailing dot normalized on both sides; an A-label is compared as the ASCII it is and never decoded, so a reference carrying a character outside ASCII is refused rather than guessed at, this toolkit shipping no IDNA implementation. An IP-ID is compared octet for octet against an `iPAddress` entry, never against a `dNSName` holding the same address as text, and no prefix or mask applies (§6.2, §6.4). Wildcards are on and `opts.wildcards: false` turns them off, which §3 requires stating either way: one wildcard, the whole of the left-most label, reaching exactly one reference label, and a presented name failing those rules is ignored with the search continuing rather than refusing the certificate (§6.3). A wildcard spanning an administrative boundary such as `*.co.uk` is out of scope as it is in §7.1, since no public-suffix list ships here. The verdict names which reference matched, so a caller can use it as the validated identity (§6.6), and lists every entry that was skipped with the reason — `match` |
 | `pki.x509` | Certificate issuance (RFC 5280 §4). `sign(spec, issuer, opts)` builds and signs a certificate from a `spec` of subject (a common-name string, an array of RDNs, or raw Name DER), the public key being certified, the validity window, an optional serial, and an optional `extensions` object. `randomSerial()` draws the same 20-octet CSPRNG serial the signer uses when the spec names none, so a serial can be reserved or logged before the certificate is issued. `parseDn(dn)` reads a distinguished-name string into `{ rdns, dn, bytes }`, the inverse of the `dn` every parsed name carries, with `bytes` being the raw Name DER `spec.subject` and `spec.issuer` take. The `issuer` is a key alone (self-signed: issuer equals subject, signed with that key), a name plus public key plus key, or an issuing certificate plus key. The signature algorithm is resolved from the signing key through the shared registry, so RSA (PKCS#1 v1.5 or PSS via `opts.pss`), ECDSA P-256/384/521, Ed25519, Ed448, ML-DSA-44/65/87, the twelve SLH-DSA sets, and the composite arms all issue without a per-algorithm branch. It encodes basic constraints, key usage, extended key usage, subject and authority key identifiers (the SKI derived by SHA-1 of the subject key; the AKI optionally naming the issuing certificate by name and serial; both emitted without being named, the SKI on every certificate and the AKI on every certificate that is not self-signed, which is what RFC 5280 §4.2.1.1 and §4.2.1.2 place on the issuing CA, and `false` declines each only where the RFC allows the omission), subject alternative names, certificate policies with their RFC 5280 section 4.2.1.4 qualifiers (a CPS pointer and a user notice), name constraints, authority information access, CRL distribution points, the freshest CRL, issuer alternative names, the policy machinery path validation acts on (policy constraints, inhibit anyPolicy, policy mappings), the RFC 3739 qualified-certificate statements including the ETSI EN 319 412-5 set, the RFC 6962 certificate-transparency pair (the precertificate poison and the embedded SCT list), the Active Directory Certificate Services enrollment extensions (certificate template, legacy template name, CA version, previous-CA-certificate hash, application policies), subject information access, subject directory attributes, and the RFC 6960 OCSP no-check responder marker, from the spec. That is every extension the shared certificate-extension decoders read, so any certificate `pki.lint.certificate` and `pki.inspect` can read in full the toolkit can also issue; any other extension is taken as pre-encoded DER. `extensions.nameConstraints` takes `{ permitted, excluded }` lists of GeneralName-form objects, restricting what the CA being issued may itself issue; each base is held to the rule for a constraint base, and the extension is emitted critical as §4.2.1.10 requires. It derives the version from the field set and enforces the serial bounds, the UTCTime/GeneralizedTime cutover, the DER default omissions, and the CA cross-field rules; a violation throws a typed `CertificateError`. Returns DER, or a PEM `CERTIFICATE` with `opts.pem`. Every arm is independently verified by OpenSSL. `extension(name, value, opts)` encodes one Extension DER from the same plain value form, for the pre-encoded array form on the certificate, CSR, and CRL signers: the criticality follows what the signer emits, `opts.critical` reaches only the extensions whose criticality is the issuer's choice, and an unregistered dotted OID takes its extnValue as bytes. Parsing stays at `pki.schema.x509.parse` — `sign`, `randomSerial`, `extension`, `parseDn` |
 | `pki.csr` | PKCS#10 certification-request issuance (RFC 2986 / RFC 2985). `sign(spec, key, opts)` builds and signs a `CertificationRequest` from a `spec` of subject (which may be empty), the public key being certified, an optional `extensionRequest` carrying the requested v3 extensions a CA copies into the issued certificate (subject alternative names, key usage, extended key usage, basic constraints, certificate policies, subject key identifier, the RFC 3739 qualified-certificate statements, the Microsoft certificate template and application policies, subject information access, subject directory attributes, or an array of pre-encoded Extension DER; an extension the issuing CA assigns, such as the authority key identifier or the OCSP no-check marker, is refused by name rather than requested), and an optional `challengePassword`. `key` (or `{ key }`) is the subject's own private key: the request is self-signed to prove possession of the private half of `subjectPublicKey`, and that proof is verified before the request is returned, which is what `openssl req -verify` checks. The signature algorithm is resolved from the subject key, so RSA (PKCS#1 v1.5 or PSS via `opts.pss`), ECDSA, EdDSA, ML-DSA, SLH-DSA, and the composite arms all sign without a per-algorithm branch. Returns DER, or a PEM `CERTIFICATE REQUEST` with `opts.pem`; malformed input throws a typed `CsrError`. `verify(request)` asks the same question of a request someone else produced, over its exact signed `certificationRequestInfo` bytes under the `subjectPKInfo` inside them, through the one path-validation signature engine. It answers `{ verified, subject, subjectPublicKeyInfo, attributes, certificationRequestInfoBytes }`, all re-derived from the signed bytes, so a CA issues from the fields the signature covers rather than from the object it passed in. A `verified: true` says the producer held that private key and that the subject and requested extensions are the ones they signed; who they are is the enrollment protocol's question, not this one. Parsing stays at `pki.schema.csr.parse` — `sign`, `verify` |

package/lib/path-validate.js

@@ -622,39 +622,118 @@ function absorbNameConstraints(state, decoded) {
 }
 
 
-function rootNode() {
-  return { depth: 0, validPolicy: OID.anyPolicy, qualifierSet: [], expectedPolicySet: [OID.anyPolicy], children: [], parent: null };
-}
-function treeWithoutParent(node) {
-  if (!node) return null;
-  return {
-    depth: node.depth,
-    validPolicy: node.validPolicy,
-    qualifierSet: node.qualifierSet,
-    expectedPolicySet: node.expectedPolicySet,
-    children: node.children.map(treeWithoutParent),
-  };
-}
-function leavesAt(tree, depth) {
+/**
+ * @internal The RFC 9618 valid_policy_graph. RFC 5280 sec. 6.1.3 step (d)(1)(i) creates one child per
+ * matching parent, so a chain where several issuer policies map to one subject policy multiplies the
+ * structure at every level: measured on a conforming chain carrying four policies mapped to each
+ * other, the tree wants 5461 nodes at six certificates against this graph's 25, and the node cap
+ * refuses the path. RFC 9618 sec. 5.2 keeps at most ONE node per (depth, policy identifier) and
+ * gives it several parents instead, which holds the same information in linear size.
+ *
+ * `state.policyIndex[depth]` maps a policy identifier to the one node at that depth, so the "is there
+ * already a node here for this policy" question every step asks is a lookup rather than a walk.
+ */
+function graphRoot() {
+  return { depth: 0, validPolicy: OID.anyPolicy, qualifierSet: [], expectedPolicySet: [OID.anyPolicy], children: [], parents: [] };
+}
+function policyLayer(state, depth) {
+  var layer = state.policyIndex[depth];
+  if (layer === undefined) { layer = intrinsic.create(null); state.policyIndex[depth] = layer; }
+  return layer;
+}
+function nodeAt(state, depth, policyOid) {
+  var layer = state.policyIndex[depth];
+  if (layer === undefined || !intrinsic.hasOwn(layer, policyOid)) return null;
+  return layer[policyOid];
+}
+function nodesAt(state, depth) {
+  var layer = state.policyIndex[depth];
+  if (layer === undefined) return [];
   var out = [];
-  if (!tree) return out;
-  (function walk(node) {
-    if (node.depth === depth) { guard.list.append(out, node); return; }
-    node.children.forEach(walk);
-  })(tree);
+  intrinsic.forEach(intrinsic.keys(layer), function (k) { guard.list.append(out, layer[k]); });
   return out;
 }
-function pruneChildless(tree, depth) {
+/** @internal One node per (depth, policy). Meeting the same pair twice joins the parents onto the
+ *  node that is already there rather than making a second one, which is the invariant the wrap-up
+ *  procedure's single-parent test depends on. */
+function addPolicyNode(state, depth, validPolicy, qualifiers, expectedPolicySet, parents) {
+  if (parents.length === 0) return null;
+  var node = nodeAt(state, depth, validPolicy);
+  if (node === null) {
+    if (state.policyNodeCount >= state.maxPolicyNodes) { state._capHit = true; return null; }
+    node = { depth: depth, validPolicy: validPolicy, qualifierSet: qualifiers ? [qualifiers] : [],
+      expectedPolicySet: expectedPolicySet, children: [], parents: [] };
+    policyLayer(state, depth)[validPolicy] = node;
+    state.policyNodeCount++;
+  }
+  for (var i = 0; i < parents.length; i++) {
+    if (node.parents.indexOf(parents[i]) !== -1) continue;
+    guard.list.append(node.parents, parents[i]);
+    guard.list.append(parents[i].children, node);
+  }
+  return node;
+}
+function removePolicyNode(state, node) {
+  node.parents.forEach(function (p) {
+    var idx = p.children.indexOf(node);
+    if (idx !== -1) p.children.splice(idx, 1);
+  });
+  node.children.forEach(function (c) {
+    var idx = c.parents.indexOf(node);
+    if (idx !== -1) c.parents.splice(idx, 1);
+  });
+  node.parents = [];
+  var layer = state.policyIndex[node.depth];
+  if (layer !== undefined && layer[node.validPolicy] === node) delete layer[node.validPolicy];
+  state.policyNodeCount--;
+}
+/** @internal RFC 9618 sec. 5.3 step (d)(3): delete every node of depth `depth` or less that has no
+ *  children, and repeat, since deleting one can leave its own parents childless. Depth zero is the
+ *  root, which is never deleted: an empty graph is a root with no children, which the caller reads
+ *  as NULL. */
+function prunePolicyChildless(state, depth) {
   for (var d = depth; d > 0; d--) {
-    leavesAt(tree, d).forEach(function (node) {
-      if (node.children.length === 0 && node.parent) {
-        var idx = node.parent.children.indexOf(node);
-        if (idx !== -1) node.parent.children.splice(idx, 1);
-      }
+    nodesAt(state, d).forEach(function (node) {
+      if (node.children.length === 0) removePolicyNode(state, node);
     });
   }
 }
-function treeIsEmpty(tree) { return tree.children.length === 0; }
+function graphIsEmpty(state) { return state.validPolicyGraph.children.length === 0; }
+/**
+ * @internal The graph a caller reads, as a flat list of nodes linked by index. A graph is not a
+ * tree, and handing back a root to walk would invite the caller to recurse into every child and
+ * rebuild the exponential structure this replaced: `JSON.stringify` on such a value re-expands a
+ * shared node once per path that reaches it. A node list is linear however it is read, holds the
+ * parent relation the wrap-up procedure needs, and carries no cycle to serialize. The root is
+ * index 0.
+ */
+function graphSnapshot(root) {
+  if (!root) return null;
+  var originals = [];
+  var indexOfNode = new intrinsic.Map();
+  (function walk(node) {
+    if (intrinsic.mapHas(indexOfNode, node)) return;
+    intrinsic.mapSet(indexOfNode, node, originals.length);
+    guard.list.append(originals, node);
+    node.children.forEach(walk);
+  })(root);
+  function idsOf(list) {
+    var out = [];
+    for (var j = 0; j < list.length; j++) {
+      if (intrinsic.mapHas(indexOfNode, list[j])) guard.list.append(out, intrinsic.mapGet(indexOfNode, list[j]));
+    }
+    return out;
+  }
+  var nodes = [];
+  for (var i = 0; i < originals.length; i++) {
+    var o = originals[i];
+    guard.list.append(nodes, {
+      id: i, depth: o.depth, validPolicy: o.validPolicy, qualifierSet: o.qualifierSet,
+      expectedPolicySet: o.expectedPolicySet, parents: idsOf(o.parents), children: idsOf(o.children),
+    });
+  }
+  return { nodes: nodes };
+}
 
 
 function isSubtreeBaseValid(tag, base) {
@@ -727,8 +806,17 @@ function _anchorSubtreeSeeds(anchor) {
 
 function initialize(certs, params, seeds, anchor) {
   var n = certs.length;
+  var state = _initialState(certs, params, seeds, anchor, n);
+  /** @internal The root is the depth-zero anyPolicy node, and it is indexed like any other, so the
+   * "is there a node at this depth for this policy" lookup answers for depth zero too. */
+  policyLayer(state, 0)[OID.anyPolicy] = state.validPolicyGraph;
+  return state;
+}
+
+function _initialState(certs, params, seeds, anchor, n) {
   return {
-    validPolicyTree: rootNode(),
+    validPolicyGraph: graphRoot(),
+    policyIndex: [],
     policyNodeCount: 1,
     maxPolicyNodes: params.maxPolicyNodes !== undefined ? params.maxPolicyNodes : constants.LIMITS.PATH_MAX_POLICY_NODES,
     minRsaModulusBits: params.minRsaModulusBits !== undefined ? params.minRsaModulusBits : constants.LIMITS.RSA_MIN_MODULUS_BITS,
@@ -757,55 +845,61 @@ function processPolicies(state, cert, i, checks) {
   try { cp = decodeExt(cert, OID.certificatePolicies); }
   catch (e) { guard.list.append(checks,{ name: "policies", ok: false, code: "path/bad-policy" }); return { fatal: true, error: e }; }
 
-  if (cp && state.validPolicyTree) {
+  if (cp && state.validPolicyGraph) {
     var policies = cp.value;
-    var depth = i - 1;
+    var prev = i - 1;
     var anyPolicyActive = state.inhibitAnyPolicy > 0 || (i < state._n && selfIssued(cert));
     var anyPolicyPresent = false;
     var anyPolicyQualifiers = null;
+
+    /** @internal RFC 9618 sec. 5.3 step (d)(1): ONE node per policy, carrying every depth i-1 node
+     * that expected it. Creating one per parent instead is what made the structure exponential. */
     policies.forEach(function (p) {
       if (p.policyIdentifier === OID.anyPolicy) { anyPolicyPresent = true; anyPolicyQualifiers = p.qualifiersBytes; return; }
-      var matched = false;
-      leavesAt(state.validPolicyTree, depth).forEach(function (node) {
-        if (contains(node.expectedPolicySet, p.policyIdentifier)) {
-          addChild(state, node, p.policyIdentifier, p.qualifiersBytes, [p.policyIdentifier], checks);
-          matched = true;
-        }
+      var parents = intrinsic.filter(nodesAt(state, prev), function (node) {
+        return contains(node.expectedPolicySet, p.policyIdentifier);
       });
-      if (!matched) {
-        leavesAt(state.validPolicyTree, depth).forEach(function (node) {
-          if (node.validPolicy === OID.anyPolicy) addChild(state, node, p.policyIdentifier, p.qualifiersBytes, [p.policyIdentifier], checks);
-        });
+      if (parents.length === 0) {
+        /** @internal (d)(1)(ii): no node expected it, so it hangs off the anyPolicy node if there
+         * is one. That single anyPolicy parent is what step (g)(2) later reads. */
+        var anyPrev = nodeAt(state, prev, OID.anyPolicy);
+        if (anyPrev === null) return;
+        parents = [anyPrev];
       }
+      addPolicyNode(state, i, p.policyIdentifier, p.qualifiersBytes, [p.policyIdentifier], parents);
     });
+
+    /** @internal (d)(2): for each policy some depth i-1 node expects, if depth i has no node for it
+     * yet, one node carrying every parent that expected it. */
     if (anyPolicyPresent && anyPolicyActive) {
-      leavesAt(state.validPolicyTree, depth).forEach(function (node) {
+      var expected = [];
+      nodesAt(state, prev).forEach(function (node) {
         node.expectedPolicySet.forEach(function (ep) {
-          var already = guard.list.anyMatches(node.children, function (ch) { return ch.validPolicy === ep; });
-          if (!already) addChild(state, node, ep, anyPolicyQualifiers, [ep], checks);
+          if (expected.indexOf(ep) === -1) guard.list.append(expected, ep);
+        });
+      });
+      expected.forEach(function (ep) {
+        if (nodeAt(state, i, ep) !== null) return;
+        var parents = intrinsic.filter(nodesAt(state, prev), function (node) {
+          return contains(node.expectedPolicySet, ep);
         });
+        addPolicyNode(state, i, ep, anyPolicyQualifiers, [ep], parents);
       });
     }
-    if (state._capHit) return { fatal: true, error: E("path/policy-tree-cap", "policy tree exceeded the node cap") };
-    pruneChildless(state.validPolicyTree, depth);
-    if (treeIsEmpty(state.validPolicyTree)) state.validPolicyTree = null;
+
+    if (state._capHit) return { fatal: true, error: E("path/policy-tree-cap", "policy graph exceeded the node cap") };
+    prunePolicyChildless(state, prev);
+    if (graphIsEmpty(state)) state.validPolicyGraph = null;
   } else if (!cp) {
-    state.validPolicyTree = null;
+    state.validPolicyGraph = null;
   }
 
-  if (!(state.explicitPolicy > 0 || state.validPolicyTree !== null)) {
+  if (!(state.explicitPolicy > 0 || state.validPolicyGraph !== null)) {
     guard.list.append(checks,{ name: "policy", ok: false, code: "path/policy-required" });
-    return { fatal: true, error: E("path/policy-required", "explicit policy required but the valid-policy tree is empty") };
+    return { fatal: true, error: E("path/policy-required", "explicit policy required but the valid-policy graph is empty") };
   }
-  return { fatal: false };
-}
-
-function addChild(state, parent, validPolicy, qualifiers, expectedPolicySet, checks) {
-  if (state.policyNodeCount >= state.maxPolicyNodes) { state._capHit = true; return; }
-  var node = { depth: parent.depth + 1, validPolicy: validPolicy, qualifierSet: qualifiers ? [qualifiers] : [], expectedPolicySet: expectedPolicySet, children: [], parent: parent };
-  guard.list.append(parent.children, node);
-  state.policyNodeCount++;
   void checks;
+  return { fatal: false };
 }
 
 function isNullOrAbsentParams(p) {
@@ -859,7 +953,7 @@ function prepareNext(state, cert, i, checks) {
   if (pm) {
     var badAny = mapsAnyPolicy(pm.value);
     if (badAny) { guard.list.append(checks,{ name: "policyMappings", ok: false, code: "path/bad-policy" }); return { fatal: true, error: E("path/bad-policy", "policyMappings must not map to or from anyPolicy (RFC 5280 6.1.4(a))") }; }
-    if (state.validPolicyTree) applyPolicyMappings(state, pm.value, i);
+    if (state.validPolicyGraph) applyPolicyMappings(state, pm.value, i);
   }
 
   state.workingIssuerName = cert.subject;
@@ -919,36 +1013,49 @@ function prepareNext(state, cert, i, checks) {
   return { fatal: false };
 }
 
+/** @internal RFC 9618 sec. 5.4 rewrites RFC 5280 sec. 6.1.4 step (b) over the graph. Each clause is
+ *  singular where the tree's was plural, because at most one node at a depth carries a given policy.
+ *
+ * A mapping naming anyPolicy on either side is skipped here as well as refused by the caller
+ * (sec. 6.1.4(a)), because mapping one in would rewrite the anyPolicy node's expected policy set and
+ * the graph's invariant is that it is always {anyPolicy}. The caller refusing it keeps that from
+ * happening today; skipping it here is what keeps it true of this function. */
 function applyPolicyMappings(state, mappings, i) {
-  var depth = i;
   if (state.policyMapping > 0) {
     var mappedFrom = intrinsic.create(null);
-    mappings.forEach(function (m) { guard.list.append(mappedFrom[m.issuerDomainPolicy] = mappedFrom[m.issuerDomainPolicy] || [], m.subjectDomainPolicy); });
-    var depthI = leavesAt(state.validPolicyTree, depth);
-    var anyNodes = depthI.filter(function (nd) { return nd.validPolicy === OID.anyPolicy; });
-    Object.keys(mappedFrom).forEach(function (idp) {
-      var idpNodes = depthI.filter(function (nd) { return nd.validPolicy === idp; });
-      if (idpNodes.length) {
-        idpNodes.forEach(function (nd) { nd.expectedPolicySet = mappedFrom[idp].slice(); });
-      } else {
-        anyNodes.forEach(function (anyNode) {
-          if (anyNode.parent) addChild(state, anyNode.parent, idp, anyNode.qualifierSet[0] || null, mappedFrom[idp].slice(), []);
-        });
-      }
+    mappings.forEach(function (m) {
+      if (m.issuerDomainPolicy === OID.anyPolicy || m.subjectDomainPolicy === OID.anyPolicy) return;
+      guard.list.append(mappedFrom[m.issuerDomainPolicy] = mappedFrom[m.issuerDomainPolicy] || [], m.subjectDomainPolicy);
     });
-  } else {
-    var mappedSet = intrinsic.create(null);
-    mappings.forEach(function (m) { mappedSet[m.issuerDomainPolicy] = true; });
-    if (!state.validPolicyTree) return;
-    leavesAt(state.validPolicyTree, depth).forEach(function (node) {
-      if (mappedSet[node.validPolicy] && node.parent) {
-        var idx = node.parent.children.indexOf(node);
-        if (idx !== -1) node.parent.children.splice(idx, 1);
+    intrinsic.forEach(intrinsic.keys(mappedFrom), function (idp) {
+      var node = nodeAt(state, i, idp);
+      if (node !== null) {
+        /** @internal (b)(1) */
+        node.expectedPolicySet = mappedFrom[idp].slice();
+        return;
       }
+      /** @internal (b)(2): no node for the issuer domain policy, so one is generated under the
+       * anyPolicy node at depth i-1, taking the depth-i anyPolicy node's qualifiers. */
+      var anyHere = nodeAt(state, i, OID.anyPolicy);
+      if (anyHere === null) return;
+      var anyPrev = nodeAt(state, i - 1, OID.anyPolicy);
+      if (anyPrev === null) return;
+      addPolicyNode(state, i, idp, anyHere.qualifierSet[0] || null, mappedFrom[idp].slice(), [anyPrev]);
     });
-    pruneChildless(state.validPolicyTree, depth - 1);
-    if (treeIsEmpty(state.validPolicyTree)) state.validPolicyTree = null;
-  }
+    return;
+  }
+  /** @internal (b)(3): mapping is inhibited, so the node for each issuer domain policy is deleted
+   * and every ancestor it leaves childless goes with it. */
+  if (!state.validPolicyGraph) return;
+  var deleted = false;
+  mappings.forEach(function (m) {
+    if (m.issuerDomainPolicy === OID.anyPolicy) return;
+    var node = nodeAt(state, i, m.issuerDomainPolicy);
+    if (node !== null) { removePolicyNode(state, node); deleted = true; }
+  });
+  if (!deleted) return;
+  prunePolicyChildless(state, i - 1);
+  if (graphIsEmpty(state)) state.validPolicyGraph = null;
 }
 
 var TARGET_UNPROCESSED_IF_CRITICAL = intrinsic.create(null);
@@ -1028,11 +1135,20 @@ function _revocationRow(row, rv) {
  * are validated at the entry point. A mis-shaped value throws `path/bad-input`,
  * so it cannot silently go unapplied. Returns `{ valid, revocationChecked, anchorConstraints,
  * path, results, workingPublicKey, workingPublicKeyAlgorithm,
- * workingPublicKeyParameters, validPolicyTree, userConstrainedPolicySet }` where `results[i].checks`
+ * workingPublicKeyParameters, validPolicyGraph, userConstrainedPolicySet }` where `results[i].checks`
  * carries a per-check reason code (`path/*`) for every step. Pure and
  * re-entrant: no input object is mutated. An empty path or a missing anchor
  * throws a typed `PathError`.
  *
+ * Certificate policies are processed as the RFC 9618 `valid_policy_graph`, which replaces the
+ * RFC 5280 sec. 6.1 `valid_policy_tree`. `validPolicyGraph` is a directed acyclic graph holding at
+ * most one node per depth and policy identifier; a node reached from several policies carries them
+ * all in `parents`, where the tree held a separate node under each. The graph is linear in the
+ * number of policies and mappings on the path, so a chain whose tree would not fit inside
+ * `maxPolicyNodes` validates. It is `{ nodes }`, a flat list linked by index, each entry
+ * `{ id, depth, validPolicy, qualifierSet, expectedPolicySet, parents, children }` where `parents`
+ * and `children` hold the ids of other entries and the root is id 0.
+ *
  * A certificate signed with the RFC 9802 stateful hash-based algorithm
  * `id-alg-hss-lms-hashsig` is verified like any other, through the toolkit's own
  * HSS engine rather than the crypto engine. The signature covers the whole
@@ -1488,28 +1604,42 @@ async function _validate(path, opts, run) {
     workingPublicKey: state.workingPublicKey,
     workingPublicKeyAlgorithm: state.workingPublicKeyAlgorithm,
     workingPublicKeyParameters: state.workingPublicKeyParameters,
-    validPolicyTree: treeWithoutParent(state.validPolicyTree),
+    validPolicyGraph: graphSnapshot(state.validPolicyGraph),
     userConstrainedPolicySet: ucps,
   });
 }
 
+/**
+ * @internal RFC 9618 sec. 5.5 rewrites RFC 5280 sec. 6.1.5 step (g) over the graph. The set is read
+ * from the nodes whose only parent is the anyPolicy node, at ANY depth rather than at the deepest
+ * one, plus the anyPolicy node at depth n. Policy qualifiers are not collected: sec. 5.5 step
+ * (g)(4)(ii) lets an application that does not use them skip it, and this returns identifiers.
+ */
 function userConstrainedPolicies(state, n) {
-  if (!state.validPolicyTree) return [];
+  if (!state.validPolicyGraph) return [];
+  var authority = [];
+  function note(node) {
+    if (authority.indexOf(node.validPolicy) === -1) guard.list.append(authority, node.validPolicy);
+  }
+  /** @internal (g)(2) */
+  for (var d = 1; d <= n; d++) {
+    nodesAt(state, d).forEach(function (node) {
+      if (node.validPolicy === OID.anyPolicy) return;
+      if (node.parents.length === 1 && node.parents[0].validPolicy === OID.anyPolicy) note(node);
+    });
+  }
+  /** @internal (g)(3) */
+  var anyAtN = nodeAt(state, n, OID.anyPolicy);
+  if (anyAtN !== null) note(anyAtN);
+
+  /** @internal (g)(5) and (g)(6) */
   var uips = state.userInitialPolicySet;
-  var anyUser = contains(uips, OID.anyPolicy);
-  var leaves = leavesAt(state.validPolicyTree, n);
-  var explicit = intrinsic.create(null), hasAnyLeaf = false;
-  leaves.forEach(function (node) {
-    if (node.validPolicy === OID.anyPolicy) hasAnyLeaf = true;
-    else explicit[node.validPolicy] = true;
-  });
-  var set = intrinsic.create(null);
-  Object.keys(explicit).forEach(function (p) { if (anyUser || contains(uips, p)) set[p] = true; });
-  if (hasAnyLeaf) {
-    if (anyUser) set[OID.anyPolicy] = true;
-    else uips.forEach(function (p) { set[p] = true; });
+  if (contains(uips, OID.anyPolicy)) return authority;
+  var out = intrinsic.filter(authority, function (p) { return contains(uips, p); });
+  if (contains(authority, OID.anyPolicy)) {
+    uips.forEach(function (p) { if (out.indexOf(p) === -1) guard.list.append(out, p); });
   }
-  return Object.keys(set);
+  return out;
 }
 
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@blamejs/pki",
-  "version": "0.8.11",
+  "version": "0.8.12",
   "description": "Pure-JavaScript PKI toolkit that owns its stack — X.509, ASN.1/DER, CMS, PQC-first.",
   "license": "Apache-2.0",
   "author": "blamejs contributors",

package/sbom.cdx.json

@@ -2,10 +2,10 @@
   "$schema": "http://cyclonedx.org/schema/bom-1.5.schema.json",
   "bomFormat": "CycloneDX",
   "specVersion": "1.5",
-  "serialNumber": "urn:uuid:41170205-6ef3-4f77-9577-7ba33e5632bc",
+  "serialNumber": "urn:uuid:d763f0be-c3b6-4191-bd7e-120bc1818e88",
   "version": 1,
   "metadata": {
-    "timestamp": "2026-09-23T22:37:26.125Z",
+    "timestamp": "2026-09-23T23:26:25.918Z",
     "lifecycles": [
       {
         "phase": "build"
@@ -19,14 +19,14 @@
       }
     ],
     "component": {
-      "bom-ref": "@blamejs/pki@0.8.11",
+      "bom-ref": "@blamejs/pki@0.8.12",
       "type": "application",
       "name": "pki",
-      "version": "0.8.11",
+      "version": "0.8.12",
       "scope": "required",
       "author": "blamejs contributors",
       "description": "Pure-JavaScript PKI toolkit that owns its stack — X.509, ASN.1/DER, CMS, PQC-first.",
-      "purl": "pkg:npm/%40blamejs/pki@0.8.11",
+      "purl": "pkg:npm/%40blamejs/pki@0.8.12",
       "properties": [],
       "externalReferences": [
         {
@@ -54,7 +54,7 @@
   "components": [],
   "dependencies": [
     {
-      "ref": "@blamejs/pki@0.8.11",
+      "ref": "@blamejs/pki@0.8.12",
       "dependsOn": []
     }
   ]