dominicnunez · Sep 12, 2026 · Sep 12, 2026
diff --git a/AGENTS.md b/AGENTS.md
@@ -2,27 +2,17 @@
 
 Agent OS is a Go modular monolith for operating persistent AI-assisted organizations.
 
-## Authoritative sources
-
-- Start with [`docs/handoff/QUICK_START.md`](docs/handoff/QUICK_START.md).
-- Treat [`docs/handoff/IMPLEMENTATION_SCOPE.yaml`](docs/handoff/IMPLEMENTATION_SCOPE.yaml) as the scope authority.
-- Use [`docs/handoff/docs/29_V1_BUILD_CONTRACT.md`](docs/handoff/docs/29_V1_BUILD_CONTRACT.md) for implementation requirements.
-- Consult the remaining preserved handoff only when the current work needs its detail.
-- Preserve `docs/handoff/` as the immutable architecture source. Publish any
-  strengthened active runtime contract outside that tree instead of silently
-  rewriting the handoff snapshot.
-- Keep repository files self-contained. Do not reference workspace-only project
-  files, local planning notes, attachments, or machine-specific project paths.
-  Cite tracked repository files or publicly accessible sources instead. Put any
-  necessary requirement directly in the appropriate repository document.
-
 ## Project-wide boundaries
 
 - Internal coordination uses runtime-owned Event Contracts over one authoritative ledger; model-generated content is untrusted.
 - Actual work determines the Task DAG and execution structure. Use model inference only where adaptive intelligence is justified.
 - Authority and completion fail closed: workers cannot expand their own capabilities or certify their own completion.
 - A2A is an external operator boundary, not internal IPC or implicit administrative authority.
-- Keep deferred architecture deferred unless a human explicitly promotes it from the handoff scope.
+- Keep deferred architecture deferred unless a human explicitly authorizes its implementation.
+- Keep repository files self-contained. Do not reference workspace-only project
+  files, local planning notes, attachments, or machine-specific project paths.
+  Cite tracked repository files or publicly accessible sources instead. Put any
+  necessary requirement directly in the appropriate repository document.
 
 Run the appropriate repository checks before a review push; hooks and CI provide
 independent enforcement rather than replacing local verification.
@@ -92,7 +82,7 @@ Follow OpenAI's [additional safety-check guidance](https://help.openai.com/en/ar
   superseded policy from an old handoff or task summary.
 - Replace contradictory active instructions at their source when adopting a
   new workflow. Do not stack competing overrides. Clearly label historical
-  records; preserve immutable handoff evidence.
+  records; preserve historical evidence.
 - Keep user-facing messages minimal. Report meaningful outcomes, blockers, and
   decisions rather than routine local commits or repeated compliance statements.
 - Use tracked project documentation to guide task work and maintain the

diff --git a/docs/EXECUTION_AUTHORITY.md b/docs/EXECUTION_AUTHORITY.md
@@ -262,14 +262,10 @@ can use exact version and purpose review, deterministic functional and authority
 checks, and authorized human judgment without held-out A/B testing. Classification
 must consider actual behavior, not a Skill's label: selection, ranking and
 tie-breaking guidance require the behavioral path below. The active security
-requirement intentionally adds this gate for discretionary Skills; the preserved
-handoff's deferral of automated evaluation does not waive it.
-Scope authorization: the project owner's requested engineering work includes
-GAP-030, "Skills must preserve declared behavioral-policy integrity," from the
-supplied `agent-os-security-gaps-complete.md`. Its explicit requirement for
-controlled held-out with-Skill/without-Skill evidence promotes that prerequisite
-for discretionary Skill activation. This active amendment records that human
-direction while preserving the handoff snapshot; it does not promote unrelated
+requirement intentionally adds this gate for discretionary Skills; deferring
+automated evaluation does not waive it. Discretionary Skill activation requires
+controlled held-out with-Skill/without-Skill evidence to preserve declared
+behavioral-policy integrity. This prerequisite does not authorize unrelated
 Lab automation or experimentation features.
 The non-discretionary classification requires independently authorized,
 version-bound evidence covering the exact content, materialization, composition

diff --git a/docs/ORGANIZATIONAL_KNOWLEDGE.md b/docs/ORGANIZATIONAL_KNOWLEDGE.md
@@ -13,8 +13,7 @@ promotion.
 The current runtime wire contract is
 [`schemas/knowledge-record.schema.json`](schemas/knowledge-record.schema.json).
 CI compares that closed schema's property and required-field sets with the Go
-`KnowledgeRecord` type. The preserved handoff schema records the earlier design
-snapshot and is not the active serialization contract.
+`KnowledgeRecord` type. This schema is the active serialization contract.
 
 ## Lifecycle
 

diff --git a/docs/development/ISO_IEC_42001_READINESS.md b/docs/development/ISO_IEC_42001_READINESS.md
@@ -47,7 +47,7 @@ Operator deployments have their own scope and evidence. Product features may sup
 | AI risks and opportunities | Threat model, adversarial tests, fail-closed authority/completion controls, and draft scoring method and initial register | PARTIAL | Review ratings and affected parties; approve treatments, owners, due conditions, and residual-risk decisions |
 | Objectives and planning | Durable Mission, Goal, Work, and Task model; evidence-backed Goal progress; ten measurable draft AIMS objectives | PARTIAL | Approve objectives, measures, owners, evaluation timing, resources, and change plans; retain operating results |
 | Resources and competence | Build contract, project guidance, CI, reviewed provider configuration, and draft competence/resource/communication procedure | PARTIAL | Approve role competence and resource criteria; perform assessments, awareness, communication, and resource review |
-| Documented information | Versioned repository, release source/provenance controls, immutable handoff, startup/recovery-verified cryptographic event-chain evidence, and a closed hash-bound public AIMS document manifest with fail-closed CI verification | PARTIAL | Approve the draft records, establish confidential evidence and external-document controls, operate retention and review, and approve an external checkpoint/retention procedure |
+| Documented information | Versioned repository, release source/provenance controls, startup/recovery-verified cryptographic event-chain evidence, and a closed hash-bound public AIMS document manifest with fail-closed CI verification | PARTIAL | Approve the draft records, establish confidential evidence and external-document controls, operate retention and review, and approve an external checkpoint/retention procedure |
 | Repository change control | Active [`Protect main`](REPOSITORY_CHANGE_CONTROL.md) ruleset requires pull requests, current-head exact CI checks, and resolved conversations; blocks deletion and force pushes; has a documented administrator recovery boundary; and has one conforming exact-head merge recorded in [PR #138](https://github.com/dominicnunez/agentos/pull/138) | PARTIAL | Monitor operation and review effectiveness. A controlled non-bypass rejection remains untested by accepted decision in closed [Issue #127](https://github.com/dominicnunez/agentos/issues/127); reopen it if a suitable identity becomes available or observed behavior conflicts with policy |
 | AI system inventory | Authenticated bounded JSON export of durable Agents, roles, lifecycle and configuration state, providers, models, runtime adapters, aggregate operations, closed projection-lifecycle Event Contract sources, explicit gaps, and an exact-byte detached SHA-256 checksum | PARTIAL | Operator-owned intended-purpose, accountable-owner, intended-user, data, dependency-risk, deployment, retention, and review records |
 | Impact assessment | Consequence candidates, user approval boundaries, and draft repeatable impact method and initial affected-party risks | PARTIAL | Approve the method; perform and retain context-specific assessments, consultation where appropriate, treatments, and residual-impact decisions |

diff --git a/docs/development/README.md b/docs/development/README.md
@@ -7,5 +7,3 @@
 - [ISO/IEC 42001 readiness](ISO_IEC_42001_READINESS.md) — continuing software controls, indefinitely deferred non-code work, preserved evidence, and certification claim boundary.
 - [Controlled AIMS information](../../governance/aims/README.md) — draft governance records, lifecycle manifest, verification, and bounded public assessment bundle.
 - [Competitive readiness](COMPETITIVE_READINESS.md) — verified product comparison and delivery priorities.
-
-The original architecture handoff remains preserved separately under [`docs/handoff/`](../handoff/README.md).
diff --git a/docs/development/V1_ACCEPTANCE_STATUS.md b/docs/development/V1_ACCEPTANCE_STATUS.md
@@ -1,8 +1,7 @@
 # V1 acceptance status
 
-This compact index maps the normative checklist in
-`docs/handoff/docs/08_IMPLEMENTATION_ROADMAP_AND_ACCEPTANCE.md` to current
-automated evidence. A passing row is architecture evidence, not authorization
+This compact index maps V1 acceptance requirements to current automated
+evidence. A passing row is architecture evidence, not authorization
 to deploy or publish.
 
 | # | Requirement | Evidence |

diff --git a/docs/handoff/AGENTS.md b/docs/handoff/AGENTS.md