WasmAgent · telleroutlook · Jul 30, 2026 · Jul 30, 2026
diff --git a/docs/15-milestones.md b/docs/15-milestones.md
@@ -2,10 +2,10 @@
 
 ## Milestone 1 — Trust Artifact Reference Implementations
 
-- [ ] `agent-trust-infra/`: Ship reference implementation for all three artifact types (AgentBOM, MCP Posture, Trust Passport)
-- [ ] `agent-trust-infra/cli`: CLI emits valid AgentBOM JSON for any agent run (`agent-trust-cli --run-id <id> --output agentbom.json`)
-- [ ] `agent-trust-infra/`: MCP Posture verification passes against sample agent manifest (`verify-posture --manifest examples/manifest.yaml`)
-- [ ] `agent-trust-infra/`: Trust Passport export includes signed AEP events (`passport export --format json --include-aep`)
+- [ ] `agentbom/`: Ship reference implementation for all three artifact types (AgentBOM, MCP Posture, Trust Passport)
+- [ ] `agentbom/cli`: CLI emits valid AgentBOM JSON for any agent run (`agentbom --run-id <id> --output agentbom.json`)
+- [ ] `agentbom/`: MCP Posture verification passes against sample agent manifest (`verify-posture --manifest examples/manifest.yaml`)
+- [ ] `agentbom/`: Trust Passport export includes signed AEP events (`passport export --format json --include-aep`)
 - [ ] `docs/`: Publish trust artifact spec docs with JSON schema validation (`docs/trust-artifacts.md` with embedded schema)
 - [ ] `tests/`: Integration test suite passes for all three artifact types (`npm run test:trust-artifacts`)
 
@@ -44,7 +44,7 @@
 - [ ] `healthcare-agent/`: Healthcare domain workload lands with HIPAA-compliant evidence collection and PHI-audit trails
 - [ ] `devops-agent/`: DevOps domain workload with deployment verification and infrastructure-as-code evidence capture
 - [ ] `finance-agent/`: Finance domain workload with SOX compliance controls and transaction-replay verification
-- [ ] `agent-trust-infra/`: Trust propagation protocol enables artifact chaining across domain boundaries (AgentBOM → AgentBOM)
+- [ ] `agentbom/`: Trust propagation protocol enables artifact chaining across domain boundaries (AgentBOM → AgentBOM)
 - [ ] `wasmagent-ops/`: Continuous verification daemon monitors running agents and alerts on trust-policy violations
 - [ ] `docs/`: Domain workload authoring guide with templates for new vertical-specific agents and verifier patterns
 - [ ] `wasmagent-js/`: Policy-as-code framework enables declarative trust rules (e.g., "require AEP events for all data mutations")

diff --git a/docs/architecture.md b/docs/architecture.md
@@ -33,9 +33,11 @@ Ingests AEP traces, applies paired-statistics checks as an evidence
 admission gate for training data, and records every training run as
 auditable evidence.
 
-### Trust artifacts — `agent-trust-infra`
+### Trust artifacts — `agentbom`
 
-Layers machine-readable identity and policy posture onto each run:
+Layers machine-readable identity and policy posture onto each run
+(`@wasmagent/agentbom-core` + `@wasmagent/agentbom-cli`; successor to the
+archived `agent-trust-infra`):
 
 - **AgentBOM** — bill of materials for an agent (model, tools, dependencies).
 - **MCP Posture** — declared and observed MCP surface and capabilities.
@@ -147,11 +149,11 @@ graph TB
 - **Evidence Store** — Immutable storage for admitted traces
 - **Training Evidence Index** — Queryable index of auditable training data
 
-### Trust artifacts layer — `agent-trust-infra`
+### Trust artifacts layer — `agentbom`
 
 ```mermaid
 graph TB
-    subgraph agent_trust_infra["agent-trust-infra"]
+    subgraph agentbom["agentbom"]
         BOM[AgentBOM Generator]
         Posture[MCP Posture Analyzer]
         Passport[Trust Passport Generator]
@@ -171,7 +173,7 @@ graph TB
 
     Artifacts[Trust Artifacts] --> Schema
 
-    style agent_trust_infra fill:#f3e5f5
+    style agentbom fill:#f3e5f5
     style Artifacts fill:#fff4e6
 ```
 
@@ -273,6 +275,6 @@ graph TB
 1. `wasmagent-js` protects a run and emits AEP events.
 2. Workloads such as `bscode` produce verifiable runtime traces.
 3. `trace-pipeline` admits and stores those traces as evidence.
-4. `agent-trust-infra` attaches AgentBOM, MCP Posture, and Trust Passport.
+4. `agentbom` attaches AgentBOM, MCP Posture, and Trust Passport.
 5. `open-agent-audit` renders the chain into audit reports.
 6. `fresharena` evaluates agents adversarially; results re-enter step 3.
diff --git a/docs/evaluation-summary.md b/docs/evaluation-summary.md
@@ -21,13 +21,13 @@ and re-usable.
   AEP evidence into `trace-pipeline`.
 - `trace-pipeline` — admits evaluation runs as evidence and records them in
   the training-audit log.
-- `agent-trust-infra` — attaches AgentBOM, MCP Posture, and Trust Passport to
+- `agentbom` — attaches AgentBOM, MCP Posture, and Trust Passport to
   each evaluated run so results bind to a specific, reproducible
   configuration.
 
 ## Feedback loop
 
 Evaluation results are not a leaderboard endpoint. They re-enter the evidence
 chain, where they inform training-data admission (`trace-pipeline`), audit
-reporting (`open-agent-audit`), and trust posture (`agent-trust-infra`). This
+reporting (`open-agent-audit`), and trust posture (`agentbom`). This
 keeps the runtime, evidence, and audit story grounded in measured behavior.
diff --git a/docs/roadmap.md b/docs/roadmap.md
@@ -20,12 +20,13 @@ maturity tiers in
 [`repository-boundaries.md`](repository-boundaries.md):
 
 - **⭐ Core** (`wasmagent-js`, `wasmagent-protocol`, `symkernel`,
-  `agent-trust-infra`, and the planned `wasmagent-py`) — runtime, protocol,
+  `agentbom`, and the planned `wasmagent-py`) — runtime, protocol,
   verification, and trust artifacts. Sustained long-term investment. AEP is
   their connective standard, sedimented from the shipping runtime and now
   versioned in `wasmagent-protocol`.
-- **🛠 Official tooling** — CLI, devtools, examples, starters. Planned; no
-  public repos yet. The tree reserves a home for them.
+- **🛠 Official tooling** — CLI, devtools, examples, starters. Landing:
+  `@wasmagent/agentbom-cli` (in `agentbom`) and `agent-golden-path` (the
+  runnable provable-stack reference app) are public; more devtools planned.
 - **🔌 Evidence surfaces** (`wasmagent-proxy`, `trace-pipeline`,
   `wasmagent-train-replay`) — extend Core to specific surfaces. Once their
   roadmaps land they move to **community maintenance**; not retired — code and
@@ -61,7 +62,10 @@ The machine-readable `focus` field in
 - 🚧 `symkernel` — Go symbolic verification backend: cel-go lightweight rules,
   wazero Wasm sandbox hard-isolation, Z3 SMT satisfiability proofs. OPA-style
   HTTP service; consumed by wasmagent-js (CelGoVerifier / Z3Verifier) and
-  future wasmagent-py via a thin adapter. Deploys on Cloudflare Containers.
+  future wasmagent-py via a thin adapter. Now imports **OPA Rego and AWS Cedar
+  policies** (translated to CEL, fail-closed — `internal/policyimport`),
+  positioning symkernel as a Z3-proof backend for existing policy languages.
+  Deploys on Cloudflare Containers.
 
 ### Workloads ✅
 
@@ -82,8 +86,11 @@ The machine-readable `focus` field in
 
 ### Trust artifacts ✅
 
-- ✅ `agent-trust-infra` — AgentBOM, MCP Posture, and Trust Passport: spec,
-  reference implementation, and CLI.
+- ✅ `agentbom` — AgentBOM, MCP Posture, and Trust Passport tooling:
+  `@wasmagent/agentbom-core` (validator, compliance checker, BOM diff/drift)
+  and `@wasmagent/agentbom-cli`. **Successor to `agent-trust-infra` (archived
+  2026-07-29)**; the library packages and CLI now live here. Trust Passport /
+  AgentBOM canonical schema → `wasmagent-protocol`.
 - 🚧 EU AI Act Annex IV compliance mapping — draft covers 20/29 Annex IV
   sub-items; deadline 2026-08-02 (Article 11 effective for high-risk AI).
 
@@ -115,7 +122,9 @@ The machine-readable `focus` field in
 
 ## Cross-cutting (planned)
 
-- 📋 Ingestion adapters: OpenTelemetry GenAI, Langfuse, LangSmith.
+- 🚧 Ingestion adapters: OpenTelemetry GenAI (field-level AEP↔OTel GenAI
+  mapping landed in `wasmagent-protocol/docs/AEP-OTEL-MAPPING.md`; adapter
+  code pending), Langfuse, LangSmith.
 - 📋 `erp-agent` public repository.
 - 📋 `wasmagent-py` Python runtime and symkernel adapter.
 - 📋 Cross-repo coherence patrol: org repo list ↔ profile README ↔

diff --git a/profile/README.md b/profile/README.md
@@ -52,7 +52,7 @@ platform's identity.
 | --- | --- |
 | [wasmagent-js](https://github.com/WasmAgent/wasmagent-js) | **Runtime** · Embedded agent runtime v1.x — WASM kernels (QuickJS, Pyodide, Wasmtime, Remote), MCP gateway + attestation, AEP emitter, capability manifests; `@wasmagent/mcp-posture` (migrated from `agent-trust-infra`), `@wasmagent/mcp-firewall`; adapters for A2A, AG-UI, AI SDK, and Claude Agent SDK. This is where signed evidence is born. |
 | [wasmagent-protocol](https://github.com/WasmAgent/wasmagent-protocol) | **Protocol** · Canonical AEP + compliance JSON Schemas — the single source of truth every repo agrees on. Published as `@wasmagent/protocol` (npm) and `wasmagent-protocol` (PyPI). |
-| [symkernel](https://github.com/WasmAgent/symkernel) | **Verification** 🚧 · Go symbolic verification backend — cel-go lightweight rules, wazero Wasm sandbox hard-isolation, Z3 SMT proofs; HTTP service consumed by wasmagent-js and wasmagent-py. "Proving it ran correctly" is the mission itself. |
+| [symkernel](https://github.com/WasmAgent/symkernel) | **Verification** 🚧 · Go symbolic verification backend — cel-go lightweight rules, wazero Wasm sandbox hard-isolation, Z3 SMT proofs; imports OPA Rego / AWS Cedar policies (translated to CEL, fail-closed) so existing policies gain formal proof. HTTP service consumed by wasmagent-js and wasmagent-py. "Proving it ran correctly" is the mission itself. |
 | [agentbom](https://github.com/WasmAgent/agentbom) | **Trust & BOM tooling** · `@wasmagent/agentbom-core`, `@wasmagent/agentbom-cli` — Agent Bill of Materials, validator, compliance checker, CLI tools. Migrated from `agent-trust-infra` (now archived). Trust Passport spec → [wasmagent-protocol](https://github.com/WasmAgent/wasmagent-protocol). |
 
 > **`wasmagent-py`** *(planned)* joins Core as the Python runtime sibling — same
@@ -111,15 +111,15 @@ The broader industry is converging on *Agent Runtime / Agent OS* as the next inf
 
 **Reliability is necessary but not sufficient.** An agent that runs stably but whose behavior cannot be verified, audited, or proven correct is not ready for production systems, regulated industries, or multi-agent trust chains. The gap between "it finished" and "it did the right thing" is exactly where WasmAgent operates.
 
-The Trusted Agent OS adds three layers that a plain Agent Runtime omits:
+Provable AI agents rest on three layers that a plain Agent Runtime omits:
 
 | Layer | What it solves | Status |
 | --- | --- | --- |
-| **Provable correctness** | `symkernel` — cel-go rules, wazero hard-isolation, Z3 SMT proofs | 🚧 in progress |
+| **Provable correctness** | `symkernel` — cel-go rules, wazero hard-isolation, Z3 SMT proofs; imports OPA Rego / AWS Cedar policies (fail-closed) so existing policies gain formal proof | 🚧 in progress |
 | **Tamper-evident evidence** | AEP (Agent Evidence Protocol) — Ed25519-signed behavioral records at gateway and runtime | shipping in `wasmagent-js` v1.x |
-| **Regulatory compliance** | EU AI Act Annex IV mapping, OWASP Agentic Top 10, NIST AI RMF | draft in `agent-trust-infra` |
+| **Regulatory compliance** | EU AI Act Annex IV mapping, OWASP Agentic Top 10, NIST AI RMF | draft in `agentbom` |
 
-The analogy: if Agent Runtime is Kubernetes (run things reliably at scale), the Trusted Agent OS is Kubernetes + audit logging + policy enforcement + compliance reporting — composable, open, and infrastructure-grade.
+Put plainly: a plain Agent Runtime runs things reliably at scale; open infrastructure for *provable* AI agents adds the proof, signed evidence, and compliance mapping on top — composable, open, and infrastructure-grade.
 
 ## Architecture