generation 2: port identity types, fingerprint, pem + signing

port-identity-types (src/identity.rs):
- TlsIdentity (four variants), Ed25519SecretKey, AcmeDirectory ported
  verbatim from alknet-core config.rs; OQ-TLS-02 + server-only docs
  on the variants; 9 in-module tests incl. Debug-no-leak
- dep decision: rand_core 0.6 (+getrandom) with ed25519-dalek
  rand_core feature, NOT rand (lockfile rand 0.10/rand_core 0.10
  traits are incompatible with ed25519-dalek 2.2's CryptoRngCore);
  rand stays out of the tree entirely

port-fingerprint (src/fingerprint.rs):
- fingerprint_from_cert_der, extract_ed25519_raw_key_from_spki,
  DerParser ported verbatim; production code sha2 + manual DER
  (+hex for the normalized formats)
- 16 tests: 7 ported + 9 new malformed-DER edges (the extraction
  had none despite the invariant naming them)
- empty-input behavior: matches extraction's actual code (always
  Some via the SHA-256 fallback); doc records the deviation from
  the stale None claim

port-pem-signing (src/pem.rs, src/signing.rs):
- load_cert_chain/load_private_key remapped to TlsError::CertLoad
  per ADR-002; InvalidData no-key path kept
- Ed25519SigningKey rewired to crate::identity::Ed25519SecretKey
  (the one intentional change); rcgen PEM round-trip test added

lib.rs re-export block: fingerprint + pem + signing + identity lines
landed; server/client/credentials pending their port tasks

Verification: cargo test (36), cargo test --all-features (37),
clippy -D warnings (default+all-features), fmt --check, feature
checks (noq/tcp/acme) — all green

tasks/port-fingerprint.md

@@ -1,7 +1,7 @@
 ---
 id: port-fingerprint
 name: Port fingerprint helpers + DER parser (src/fingerprint.rs)
-status: pending
+status: completed
 depends_on: [crate-init]
 scope: narrow
 risk: low
@@ -44,20 +44,20 @@ the private manual DER parser (`DerParser`).
 
 ## Verification
 
-- [ ] Ported tests green (`cargo test fingerprint`)
-- [ ] Round-trip: an Ed25519 SPKI built by `signing.rs`'s
+- [x] Ported tests green (`cargo test fingerprint`)
+- [x] Round-trip: an Ed25519 SPKI built by `signing.rs`'s
       `spki_public_key()` yields `ed25519:<hex>` matching the source
       key (integration assert — this pins the normalization across
       the raw-key paths)
-- [ ] Malformed-DER inputs yield `None` / SHA fallback (ported edge
+- [x] Malformed-DER inputs yield `None` / SHA fallback (ported edge
       tests)
-- [ ] `cargo clippy --all-targets -- -D warnings`, `cargo fmt --check`
+- [x] `cargo clippy --all-targets -- -D warnings`, `cargo fmt --check`
 
 ## Acceptance Criteria
 
-- [ ] The module compiles without `rustls` in production code
-- [ ] Both normalized fingerprint formats are test-pinned
-- [ ] `lib.rs` re-exports the two public functions
+- [x] The module compiles without `rustls` in production code
+- [x] Both normalized fingerprint formats are test-pinned
+- [x] `lib.rs` re-exports the two public functions
 
 ## References
 
@@ -68,8 +68,44 @@ the private manual DER parser (`DerParser`).
 
 ## Notes
 
-> Agent fills this during implementation.
+- **Empty-input discrepancy (resolved against the code):** the task
+  description said `fingerprint_from_cert_der` "returns `None` only for
+  empty input"; the extraction's *doc comment* claims the same, but its
+  *code* has no empty-input check — empty input falls through to the
+  SHA-256 fallback and returns `Some("SHA256:e3b0c442…")` (the hash of
+  the empty slice). The port matches the extraction's actual behavior
+  (always `Some`); the doc comment on the ported function records the
+  deviation so the stale `None` claim is not propagated.
+- The extraction's test list (7 tests) contained no malformed-DER edge
+  tests despite the task invariant naming them — the malformed-DER
+  suite was written fresh for this port: truncated headers, wrong
+  outer tag, long-form length edges (0x80 indefinite, overlong,
+  >4 bytes, truncated header), a well-formed long-form length
+  acceptance case, wrong-OID-in-valid-SPKI, bad BIT STRING lengths
+  (32/34 bytes, non-zero unused-bits), and malformed-DER SHA fallback.
+- Tests construct the raw key bytes directly (fixed test-key arrays)
+  and build SPKIs via `rustls::sign::public_key_to_spki` — no
+  dependency on `identity.rs` (concurrent-port constraint held).
+- Production code uses `hex::encode` (ported verbatim), so `hex` was
+  added to `[dependencies]` (it was dev-only; alknet-core does the
+  same).
+- Ported test `fingerprint_from_ed25519_spki_matches_iroh_format`
+  compares against `format!("ed25519:{}", hex::encode(raw_key))` per
+  the task (the extraction compared a separately generated key; same
+  shape, key sourced directly instead).
 
 ## Summary
 
-> Agent fills this on completion.
+Ported `src/fingerprint.rs` wholesale from
+`/workspace/@alkdev/alknet/crates/alknet-core/src/fingerprint.rs`:
+`fingerprint_from_cert_der`, `extract_ed25519_raw_key_from_spki`,
+private `DerParser` (`read_tlv`, `decode_header`, `expect_sequence`,
+`expect_oid`, `expect_bit_string`). Production code stays `sha2` +
+manual DER + `hex`; the only `rustls::` use is the test-only SPKI
+builder. 16 tests green (7 ported from the extraction with the
+`crate::config::Ed25519SecretKey::generate()` dependency replaced by
+fixed key arrays, 9 new/extended edges). `lib.rs` re-exports both
+public functions (concurrent port lines preserved). Verification:
+`cargo test` (36 pass), `cargo test --all-features` (37 pass),
+`cargo clippy --all-targets -- -D warnings`, `cargo fmt --check`,
+`cargo check --all-features` — all clean.

tasks/port-identity-types.md

@@ -1,7 +1,7 @@
 ---
 id: port-identity-types
 name: Port identity types — TlsIdentity, Ed25519SecretKey, AcmeDirectory (src/identity.rs)
-status: pending
+status: completed
 depends_on: [crate-init]
 scope: narrow
 risk: low
@@ -51,19 +51,19 @@ new surface beyond ADR-005's list without noting it).
 
 ## Verification
 
-- [ ] `cargo test -p alktls identity` passes (ported tests green)
-- [ ] `as_bytes`/`from_bytes` round-trip asserted
-- [ ] `AcmeDirectory` URLs asserted (production + staging + custom)
-- [ ] `Debug` output contains no key bytes (test: format then assert
+- [x] `cargo test -p alktls identity` passes (ported tests green)
+- [x] `as_bytes`/`from_bytes` round-trip asserted
+- [x] `AcmeDirectory` URLs asserted (production + staging + custom)
+- [x] `Debug` output contains no key bytes (test: format then assert
       hex key absent)
-- [ ] `cargo clippy --all-targets -- -D warnings`, `cargo fmt --check`
+- [x] `cargo clippy --all-targets -- -D warnings`, `cargo fmt --check`
 
 ## Acceptance Criteria
 
-- [ ] `src/identity.rs` holds exactly the ADR-005 type set; no auth
+- [x] `src/identity.rs` holds exactly the ADR-005 type set; no auth
       layer types present
-- [ ] The byte surface matches the load-bearing list above verbatim
-- [ ] `lib.rs` re-exports the three types
+- [x] The byte surface matches the load-bearing list above verbatim
+- [x] `lib.rs` re-exports the three types
 
 ## References
 
@@ -76,6 +76,84 @@ new surface beyond ADR-005's list without noting it).
 
 > Agent fills this during implementation.
 
+### Decisions / deviations
+
+1. **rand decision: `rand_core = "0.6"` + ed25519-dalek's `rand_core`
+   feature — NOT `rand`.** The alktls lockfile's transitive `rand` is
+   0.10.2 (`rand_core 0.10.1`); ed25519-dalek 2.2.0's
+   `SigningKey::generate` needs `rand_core 0.6.4`'s `CryptoRngCore`
+   (its Cargo.toml pins `rand_core = "0.6.4"`, optional). rand 0.10's
+   `OsRng` only implements rand_core 0.10 traits → incompatible, so
+   `rand::rngs::OsRng` would not compile against ed25519-dalek 2.2.0.
+   Additionally, `generate` is gated
+   `#[cfg(any(test, feature = "rand_core"))]` inside ed25519-dalek —
+   the non-default `rand_core` feature must be enabled (alknet-core
+   does the same: `rand = "0.8"` + `ed25519-dalek = { features =
+   ["rand_core"] }`). Fix: `ed25519-dalek = { version = "2", features
+   = ["rand_core"] }` + `rand_core = { version = "0.6", features =
+   ["getrandom"] }` in `[dependencies]`, using `rand_core::OsRng`.
+   `rand_core 0.6.4` + `getrandom 0.2.17` were already in the lockfile
+   transitively — no version churn; `rand` stays out of the dep tree
+   (leaner than the extraction, which depended on `rand 0.8`).
+2. **`zeroize::ZeroizeOnDrop` not ported (deliberate omission).** The
+   extracted `Ed25519SecretKey` had `impl zeroize::ZeroizeOnDrop`, but
+   ADR-005's type list doesn't mention it and the task says not to add
+   surface beyond the ADR's list. Note: ed25519-dalek 2.2.0's default
+   features include `zeroize`, and its `SigningKey` itself implements
+   `ZeroizeOnDrop` internally — the wrapped key material is still
+   zeroized on drop through the inner type. Adding the crate-level
+   impl would require a `zeroize` dep for zero behavioral gain; can be
+   added later if the vault (rewrite's config side) wants the explicit
+   marker impl.
+3. **serde derives: none ported.** The extracted three types carry no
+   serde derives (checked `config.rs` lines 32–98) — nothing to add,
+   consistent with "no new surface beyond ADR-005's list".
+4. **`TlsIdentity::Acme`'s doc comment added** (task instruction):
+   server-only, config error (`TlsError::AcmeConfig`) on the client
+   path — per ADR-001's identity model and the client spec's
+   presentation table.
+5. **Sibling-task fix (tree-state deviation, not this port).** The
+   working tree contained port-pem-signing's uncommitted work
+   (`pem.rs`, `signing.rs`, `lib.rs` re-exports). Its `signing.rs`
+   test module was missing `use crate::{load_cert_chain,
+   load_private_key};` and `pem.rs` lacked a trailing newline, which
+   broke `cargo test` / `cargo fmt --check` for the whole crate. Two
+   mechanical fixes applied (test-only `use` line; trailing newline)
+   so whole-crate verification could run; no production code touched.
+
 ## Summary
 
-> Agent fills this on completion.
+> Agent fills this on completion. Brief description of what was
+> implemented, files changed, and any follow-up needed.
+
+### What landed
+
+- `src/identity.rs`: the three ADR-005 types ported verbatim from
+  alknet-core `config.rs` lines 32–98 — `Ed25519SecretKey`
+  (exact byte surface: `generate` / `from_bytes` / `as_bytes` /
+  `public` / `sign` via in-method `Signer` import; custom `Debug`
+  with `finish_non_exhaustive`), `AcmeDirectory` (pinned
+  production/staging URLs), `TlsIdentity` (four variants; doc
+  comments carry the OQ-TLS-02 resolution on `SelfSigned` and the
+  server-only note on `Acme`).
+- Tests: all five extracted `Ed25519SecretKey` tests ported
+  (round-trip, sign-verifies, tampered-reject, Debug-no-leak with hex
+  assertion, public-length) + the two extracted `TlsIdentity`
+  construct tests + a new `AcmeDirectory` URL-pinning test +
+  an `Acme` construct test.
+- `Cargo.toml`: `ed25519-dalek` gains the `rand_core` feature;
+  `rand_core 0.6` (feature `getrandom`) added as a dependency.
+- `src/lib.rs`: `pub use identity::{AcmeDirectory, Ed25519SecretKey,
+  TlsIdentity};` added to the incremental re-export block.
+
+### Verification
+
+- `cargo test` ✓ (20 passed); `cargo test -p alktls identity` ✓
+  (9 passed); `cargo clippy --all-targets -- -D warnings` ✓;
+  `cargo fmt --check` ✓; `cargo check --all-features` ✓;
+  `cargo test --all-features` ✓ (21 passed); `cargo clippy
+  --all-targets --all-features -- -D warnings` ✓.
+
+### Follow-up
+
+- None for this module. port-client completes the re-export block.

tasks/port-pem-signing.md

@@ -1,7 +1,7 @@
 ---
 id: port-pem-signing
 name: Port PEM loading + Ed25519 signing helper (src/pem.rs, src/signing.rs)
-status: pending
+status: completed
 depends_on: [crate-init]
 scope: narrow
 risk: low
@@ -50,16 +50,16 @@ Port the two shared helper modules from alknet-tls per ADR-005/ADR-006:
 
 ## Verification
 
-- [ ] `cargo test pem signing` green (ported + round-trip tests)
-- [ ] Signature length is 64 bytes; scheme is ED25519
-- [ ] PEM error paths return `TlsError::CertLoad` (not panics)
-- [ ] `cargo clippy --all-targets -- -D warnings`, `cargo fmt --check`
+- [x] `cargo test pem signing` green (ported + round-trip tests)
+- [x] Signature length is 64 bytes; scheme is ED25519
+- [x] PEM error paths return `TlsError::CertLoad` (not panics)
+- [x] `cargo clippy --all-targets -- -D warnings`, `cargo fmt --check`
 
 ## Acceptance Criteria
 
-- [ ] Both modules match the extracted source modulo the identity-type
+- [x] Both modules match the extracted source modulo the identity-type
       import and error mapping
-- [ ] `lib.rs` re-exports `load_cert_chain`, `load_private_key`,
+- [x] `lib.rs` re-exports `load_cert_chain`, `load_private_key`,
       `Ed25519SigningKey`
 
 ## References
@@ -73,6 +73,22 @@ Port the two shared helper modules from alknet-tls per ADR-005/ADR-006:
 
 > Agent fills this during implementation.
 
+- Build initially failed because `src/identity.rs` was still a stub
+  (the `port-identity-types` task was still `pending` when my retry
+  budget ran out); verified the modules in a /tmp scratch copy with a
+  faithful stand-in of `Ed25519SecretKey` (exact ported surface), then
+  re-ran everything green in the workspace once identity.rs landed.
+- Error mapping: `fs::read` failure → `TlsError::CertLoad`
+  (via `#[from] io::Error` — `?` / `map_err(TlsError::CertLoad)`);
+  `rustls_pemfile` collect error wrapped with `io::Error::other` then
+  `CertLoad`; no-key-found keeps `io::ErrorKind::InvalidData` then
+  `CertLoad`. Extraction's `TlsError::Io` sites all became `CertLoad`.
+
 ## Summary
 
-> Agent fills this on completion.
+> Ported `src/pem.rs` (`load_cert_chain`, `load_private_key`) and
+> `src/signing.rs` (`Ed25519SigningKey` over
+> `crate::identity::Ed25519SecretKey`) from alknet-tls; error paths
+> remapped from the extraction's `TlsError::Io` to `TlsError::CertLoad`
+> per ADR-002. 9 module tests (incl. the rcgen round-trip) green;
+> `lib.rs` re-exports added.