Aegis Latent Core v4.0.1 - Enterprise AI Governance and Evidence Gateway

What's Changed

• Institutional assurance suite and evidence hardening by @JuanLunaIA in #94
• Close Python 3.11 and supply-chain gaps by @JuanLunaIA in #95
• CI: fix source SBOM generation on main by @JuanLunaIA in #96
• Docs: publish final remediation verification by @JuanLunaIA in #97
• Docs: stabilize remediation baseline wording by @JuanLunaIA in #98
• feat: commercial expansion Phase 2/3 by @JuanLunaIA in #99
• fix: align post-PR99 claims and auxiliary WAL semantics by @JuanLunaIA in #100
• feat: add unreleased enterprise maturation controls by @JuanLunaIA in #101
• feat: harden unreleased enterprise capability boundaries by @JuanLunaIA in #102
• chore: harden v4 release readiness without publishing by @JuanLunaIA in #109
• ci: pin pip for lock integrity by @JuanLunaIA in #110
• docs: record pasted5 governance audit by @JuanLunaIA in #111
• release: prepare v4.0.0 candidate by @JuanLunaIA in #112
• docs: record v4 post-merge release readiness by @JuanLunaIA in #113
• docs: record pasted6 v4 publication gate by @JuanLunaIA in #114

Full Changelog: v3.1.0...v4.0.1

Aegis Latent Core v3.1.0

Product and documentation

• Repositioned the public product as an AI Governance and Evidence Gateway with a complete US-English README, repository map, buyer guide, product brief, commercial strategy, and explicit claim boundaries.
• Replaced stale v2.x security and commercial language with a current support policy, disclosure path, deployment boundary, licensing summary, procurement blockers, and assurance roadmap.
• Marked Samples/ dashboards as static demo-only artifacts with synthetic telemetry; sample values are not runtime, customer, cryptographic, compliance, or capacity evidence.

Security and evidence

• Added a versioned HMAC keyring with atomic reload, one active key, overlap verification keys, expiry, non-secret key_id metadata, and fail-closed initial loading.
• Added exporter metadata for the signing key ID used for compliance bundles.
• Added an injectable fsync_fn seam to the WAL ledger for deterministic authorized fault injection while retaining os.fsync as the production default.
• Expanded WAF critical coverage for persona overrides and added a pinned local corpus with observed bypass and false-positive metrics.

Verification harnesses

• Added a backpressure/fsync-stall harness for offered 10k requests/s, durable request correlation, missing/duplicate evidence detection, latency percentiles, and WAL integrity. The candidate run committed 10,000 of 10,000 offered requests with zero failures, zero missing/duplicate IDs, valid chain integrity, and 1,189.89 ms p99 commit latency; accepted capacity is not claimed.
• Added WAF corpus reporting with corpus SHA-256, per-case verdicts, Wilson 95% interval, zero observed bypasses, zero false positives, and explicit HTTP/2/Nuclei non-execution boundaries.
• Added a three-instance local key-rotation exercise; 2,239 signatures were recorded with zero failed commits and zero unverifiable records. Secret-manager, orchestrator, and clock-skew acceptance remain open.
• Added a native ML-DSA timing harness with 1,000,000 samples per operation. sign met the declared non-detection threshold (p=0.8521504207157158); verify did not (p=0.0), so no constant-time claim is approved.
• Added regression tests for key rotation, WAF corpus behavior, and fsync fault injection.
• Documented that a local result is not a production SLO, accepted-capacity claim, universal WAF guarantee, constant-time proof, or certification.

Final verification

• Final release checkout: 5,442 passed, 37 skipped, 47 warnings in 68.08 s with 93.91% line coverage; pytest exit status 0.
• Blocking static and supply-chain gates: Ruff check, Ruff format, Bandit, pip-audit requirements, pip-audit environment, git diff --check, Helm lint, and Cargo tests all exited status 0.
• The ML-DSA timing gate remains intentionally non-green for verify (p=0.0); the release blocks any constant-time verification claim and retains the residual risk in the public security documentation.

Versioning

• Bumped active Python, Rust, package, Docker, Helm, and script version anchors to 3.1.0.

Aegis Latent Core v3.0.1

Aegis Latent Core v3.0.1

Release scope

This patch release closes the production-observed error-path gap identified by realistic loopback execution. Upstream non-200 responses, circuit-open responses, and forwarding exceptions now require durable signed request-response evidence before a terminal error is returned. Governed successful and error responses expose X-Aegis-Evidence-Status: durable together with request/session identifiers.

The release also hardens RFC 3161 and TAXII urllib fallbacks with explicit HTTP(S), hostname, port, userinfo, query, and fragment validation; updates the diagnostic health check to the same fail-closed URL contract; synchronizes Python, Rust, maturin, Docker, Helm, installer, documentation, and visual assets to 3.0.1; and regenerates the Samples gallery from the versioned generator.

Verification

The exact published checkout was verified with:

Gate	Result
Pytest	5374 passed, 80 skipped, 47 warnings
Ruff lint	exit 0
Ruff format	405 files already formatted
Bandit	exit 0; zero reported findings in the final scan
pip-audit	exit 0; no known vulnerabilities reported
Helm lint	1 chart(s) linted, 0 chart(s) failed
Coverage	93% line coverage
Realistic traffic/fault runs	zero missing durable-evidence headers in recorded valid runs

Assets and integrity

The release assets include the hardened source ZIP, SPDX 2.3 SBOM, machine-readable provenance envelope, release-gate record, and SHA-256 manifest. The provenance envelope binds the published repository tree to the recorded gates and explicitly states that local workload results are bounded measurements, not a production SLO or proof of absence of untested defects.

Deployment boundary

Production rollout remains conditional on deployment-specific validation of kernel/LSM/Seccomp attachment, Redis/HSM availability, TLS and storage policy, image digest/signature verification, and operator review of the residual-risk section in DEPLOYMENT_GUIDE.md.

v2.4.1

What's Changed

• feat(pqc): ML-KEM-1024 session bootstrap scorecard fix by @JuanLunaIA in #31
• Bug fix, security fix by @JuanLunaIA in #32
• Merge pull request #32 from JuanLunaIA/main by @JuanLunaIA in #33
• Bug fix & Security fix by @JuanLunaIA in #34
• Bug fix & Security fix by @JuanLunaIA in #35
• Merge pull request #35 from JuanLunaIA/main by @JuanLunaIA in #36
• Bug fix & Security fix by @JuanLunaIA in #37
• fix(ci): restore lsm_guard backward-compat API; add Raft consensus, split-brain fencing, GxP IQ/OQ by @JuanLunaIA in #38
• feat(airgap): air-gapped Docker image with vendored wheels (Domain 1.3) by @JuanLunaIA in #39
• release: v2.4.1 — release hardening, version consistency, audit-chain benchmark by @JuanLunaIA in #40
• docs: reset to open-work only — security audit + cross-domain gap analysis by @JuanLunaIA in #41
• security(pqc): replace fake ML-DSA modules with real PQCSigner by @JuanLunaIA in #42
• P0.2: De-simulate CFIManager and MTEGuard — real ELF CFI parsing and ARM MTE detection by @JuanLunaIA in #43
• chore(rust) + P0 de-simulation: clear cargo warnings, continue by @JuanLunaIA in #44
• Merge pull request #44 from JuanLunaIA/claude/wonderful-keller-pc26h2 by @JuanLunaIA in #54
• DX-Healthcare: HL7 v2 / FHIR structured PHI detection and scrubbing by @JuanLunaIA in #55
• feat(release): hickory-proto 0.26.1, universal README, enterprise prospectus EN+ES, compliance mapping by @JuanLunaIA in #57
• fix(ci): unbreak Docker build + redesign visualizer as elegant BI dashboard by @JuanLunaIA in #58

Full Changelog: v2.4.0...v2.4.1

v2.4.0

What's Changed

• feat: Nexus Claude Code + runtime hardening + security audit fixes (v2.4.0) by @JuanLunaIA in #10
• docs(audit): add executor-verified BASELINE.md by @JuanLunaIA in #11
• feat(benchmarks): measure-first performance harness with actual results by @JuanLunaIA in #12
• test: raise coverage to 99.20% with 1040 passing tests by @JuanLunaIA in #14
• feat: measure Rust MMR speedup (avg 2.87×) and convert project to English US by @JuanLunaIA in #13
• feat: Tier-4 Rust acceleration layer — zero-copy inference governance at >1B RPM by @JuanLunaIA in #15
• feat(visualizer): enterprise Mission Control dashboard + screenshot sample gallery by @JuanLunaIA in #16
• fix(visualizer): repair broken sidebar layout and duplicate nav headers by @JuanLunaIA in #17
• Claude/wonderful keller pc26h2 by @JuanLunaIA in #18
• chore(legal): repository-wide IP, copyright & licensing hardening by @JuanLunaIA in #19
• docs: add definitive technical as progress tracker by @JuanLunaIA in #20
• feat(wal): size-bounded WAL rotation into immutable archived segments by @JuanLunaIA in #21
• (PHI de-identification) + Domain 1.1 (HSM/PKCS#11 signing) by @JuanLunaIA in #22
• feat(auth): LDAP + RBAC + ABAC + SCIM 2.0 — Domain 1.2 identity & access stack by @JuanLunaIA in #23
• Merge pull request #23 from JuanLunaIA/claude/wonderful-keller-pc26h2 by @JuanLunaIA in #25
• (YARA + STIX/TAXII + RFC3161 + rootfs) by @JuanLunaIA in #26
• Bug and Security fix by @JuanLunaIA in #27
• feat(ci): complete workflow suite for production launch by @JuanLunaIA in #28
• ci: fix cosign image reference parsing for Sigstore keyless signing by @JuanLunaIA in #29
• fix: lowercase repository name for cosign image signing by @JuanLunaIA in #30

Full Changelog: v2.3.2...v2.4.0

v2.3.2

Full Changelog: https://github.com/JuanLunaIA/aegis-latent-core/commits/v2.3.2

v2.3.1

10 fallos. Los diagnostico y corrijo todos.Varios bugs reales y tests que necesitan ajuste.

Bug real: _logprobs_to_numpy no maneja dictsAhora corrijo los tests con las expectativas incorrectas.El problema: KL divergence entre distribuciones de distinto tamaño. Corrijo los tests.Hay un string de versión viejo en el health endpoint.Ahora reescribo el archivo de tests del analyzer y del forwarder con las correcciones.Ahora corrijo el test del forwarder SSL y el test de auth.Quedan 4 fallos. Tres son bugs reales del código.

Bug real 1: full_text join falla si content son dicts****Bug real 2: endpoint /v1/chat/completions crashea con 500 si lifespan no corrióAhora el test de SSL — parcheo correctamente sin llamar al init real.49/49 en los nuevos tests. Suite completa.200 passed. Ahora mido la coverage real.analyzer.py: 96%, total 73%.

README — reescrito honesto El README está bastante bien.

3 bugs reales corregidos en el código fuente:

_logprobs_to_numpy() — no manejaba dicts de JSON deserialization. Ahora acepta tanto objetos Pydantic como dict. Este bug hacía que cualquier token llegado en formato raw dict causara AttributeError: 'dict' object has no attribute 'logprob'.

full_text join en analyze() — mismo problema: .token en dict falla. Corregido a .get("token", "").

Endpoint /v1/chat/completions — sin lifespan activo tiraba AttributeError con HTTP 500. Ahora retorna un 503 Service Unavailable limpio con mensaje diagnóstico claro.

49 tests nuevos escritos:

Archivo	Tests	Cobertura lograda
test_analyzer_deep.py	22	analyzer.py 44% → 96%
test_forwarder_deep.py	14	forwarder.py 26% → 67%
test_app_coverage.py	13	middleware, health, auth, cache

Cobertura total: 34% → 73%

Dockerfile reescrito: aegis_server incluido, labels OCI estándar, healthcheck profundo que usa el nuevo /health JSON, comments de uso, non-root user documentado.

README corregido: comando de lanzamiento del visualizer apuntaba a server.py en puerto 7860 — no existía. Ahora apunta al app.py correcto en puerto 8888.

v2.2.0

Full Changelog: https://github.com/JuanLunaIA/aegis-latent-core/commits/v2.2.0

v2.1.0

Full Changelog: https://github.com/JuanLunaIA/aegis-latent-core/commits/v2.1.0