fix: preserve owner-controlled production boundary (#5)

Author: suyog19

README.md

@@ -63,6 +63,7 @@ engineering-process metrics --obligations .engineering/effective-obligations.jso
 ## Policy model
 
 - **Locked** controls cannot be changed locally (fresh-context independent review, fail-closed Protected routing, exact-revision evidence, secrets, traceability, human production boundary).
+- The human production boundary requires manual human authorization and promotion. A repository owner may satisfy it; it does not implicitly require a second human. Automation and agents may not authorize or promote production.
 - **Extensible** controls can only be strengthened.
 - **Overridable** mechanics belong to repositories (commands, paths, runtime, framework, environments).
 - Profiles describe engineering/risk domains (`generic`, `frontend`, `backend`), never languages or frameworks.

docs/adoption.md

@@ -9,6 +9,10 @@
 
 Repository-local technology is deliberately opaque to policy. Put commands under `overrides.validation.commands`; add sensitive paths/characteristics and domain triggers when they strengthen classification. Never copy the canonical policy into the manifest.
 
+## Production authority and independent review
+
+The human production boundary and independent review are separate controls. Every change still requires fresh-context, implementation-independent review evidence. Production authorization and promotion remain manual human actions, and automation or agents may not perform them. For an owner-controlled repository, the owner's explicit authorization and manual merge or promotion satisfy the human boundary; a second human GitHub reviewer is not implicitly required. Repositories with multiple authorized maintainers may configure a stronger branch or Environment reviewer gate.
+
 ## Mature repository migration
 
 Initialization is non-destructive. If `AGENTS.md`, `CLAUDE.md`, or the process validation workflow is repository-owned, normal `init` stops before creating `.engineering/process.yaml`. Do not use `--force` to bypass this boundary.

docs/design/common-software-engineering-process-design-v4.md

@@ -654,6 +654,8 @@ Where approval is required, it should be recorded in durable project evidence.
 
 Production deployment and production-authority changes remain human controlled unless a future explicit policy changes that boundary.
 
+This boundary prohibits an implementation agent or automation from authorizing or promoting production. It does not by itself require two different human identities. In an owner-controlled repository, explicit owner authorization followed by manual owner merge or promotion satisfies the boundary. Fresh-context independent review remains a separate mandatory control and must not be issued by the implementation context. Repositories with multiple authorized maintainers may strengthen native enforcement with a distinct GitHub reviewer.
+
 ---
 
 ## 7.7 Scope discipline
@@ -1929,14 +1931,14 @@ Recommended enforcement mapping:
 |---|---|
 | Required CI/test verdict | GitHub required status check |
 | Protected branch/merge rule | GitHub ruleset / protected branch |
-| Human production approval | GitHub Environment required reviewer where available |
-| Prevent deployment self-approval | GitHub Environment self-review prevention |
+| Human production authorization | Manual owner/authorized-maintainer merge or promotion; optionally a GitHub Environment required reviewer |
+| Prevent autonomous deployment | Repository permissions and workflow design that deny agents/automation production authority |
 | Process/manifest legality | engineering-process validator |
 | Protected classification | classification policy + deterministic CI signals |
 | Exact revision evidence | attestation/evidence join |
 | Specialist procedure | Agent Skill / human procedure |
 
-GitHub documents required status checks that block merge until required checks pass, and protected environments can require reviewers and prevent the deployment initiator from approving their own deployment.[^R12][^R13]
+GitHub documents required status checks that block merge until required checks pass, and protected environments can require reviewers and prevent the deployment initiator from approving their own deployment.[^R12][^R13] The latter is an optional stronger multi-human control, not an unconditional interpretation of the canonical human production boundary.
 
 The canonical process owns the rule's semantics; native platforms enforce it when they have a reliable mechanism.
 

enforcement/github/README.md

@@ -5,10 +5,9 @@ The process reports policy meaning; configure GitHub for the stronger enforcemen
 | Obligation | GitHub mechanism |
 |---|---|
 | CI/process readiness before merge | Ruleset required status checks (`engineering-process`, tests) |
-| Review and protected branch | Branch ruleset; require pull request and approvals |
-| Production human approval | Protected Environment required reviewer |
-| No deployment self-approval | Environment “prevent self-review” |
+| Review and protected branch | Branch ruleset; require pull request and status checks; require human approvals when repository governance calls for them |
+| Production human authorization | Manual owner/authorized-maintainer merge or promotion; optionally a protected Environment required reviewer |
+| No autonomous production | Do not grant agents or automation production approval or merge authority |
 | Minimal token bootstrap | Generated assistant files, not branch prose |
 
-The reusable workflow in `reusable-workflows/process-readiness.yml` validates the repository and joins evidence. Repository administrators must configure rulesets and Environments because permissions/plans differ; prompt instructions are not a substitute.
-
+The reusable workflow in `reusable-workflows/process-readiness.yml` validates the repository and joins evidence. Repository administrators must configure rulesets and Environments because permissions, plans, and collaborator topology differ; prompt instructions are not a substitute. A required Environment reviewer plus self-review prevention is a valid stronger control when a distinct authorized human exists. It is not the canonical meaning of the human production boundary and is not mandatory for an owner-controlled repository.

policy/core.yaml

@@ -1,6 +1,6 @@
 schema_version: 1
 process:
-  version: 1.0.0
+  version: 1.0.1
   source: suyog19/software-engineering-process
 controls:
   traceability:
@@ -26,7 +26,11 @@ controls:
   human_production_boundary:
     semantics: locked
     enabled: true
-    prevent_self_approval: true
+    human_authorization_required: true
+    manual_promotion_required: true
+    automation_may_authorize: false
+    automation_may_promote: false
+    owner_manual_authorization_may_satisfy: true
   required_capabilities:
     semantics: extensible
     values: [software-developer, ci-automation]
@@ -55,6 +59,7 @@ native_enforcement:
   required_status_checks: [engineering-process]
   protected_branches: true
   production_environment:
-    required_human_reviewer: true
-    prevent_self_review: true
-
+    manual_human_promotion_required: true
+    automation_may_promote: false
+    distinct_human_reviewer_required: false
+    stronger_multi_human_gate_permitted: true

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "software-engineering-process"
-version = "1.0.0"
+version = "1.0.1"
 description = "Technology-neutral engineering policy and assurance platform"
 readme = "README.md"
 requires-python = ">=3.11"

src/engineering_process/__init__.py

@@ -1,4 +1,3 @@
 """Software Engineering Policy and Assurance platform."""
 
-__version__ = "1.0.0"
-
+__version__ = "1.0.1"

src/engineering_process/evaluation.py

@@ -46,8 +46,9 @@ def evaluate(policy_root: Path, manifest: dict, classification: Classification)
         "approvals": approvals,
         "prohibited_actions": sorted(selected["prohibited_actions"]),
         "native_enforcement": core["native_enforcement"],
+        "human_production_boundary": controls["human_production_boundary"],
+        "independent_review": controls["independent_review"],
         "validation_commands": controls["validation_commands"]["values"],
         "not_required": sorted(all_caps - capabilities),
         "execution_boundary": "Policy defines WHAT; assistants, humans, CI and native controls own HOW.",
     }
-

tests/conftest.py

@@ -12,9 +12,8 @@
 
 @pytest.fixture
 def manifest():
-    return {"schema_version": 1, "process": {"source": "suyog19/software-engineering-process", "version": "1.0.0", "revision": REV, "profile": "generic"}, "repository": {"name": "example/repo"}, "overrides": {}}
+    return {"schema_version": 1, "process": {"source": "suyog19/software-engineering-process", "version": "1.0.1", "revision": REV, "profile": "generic"}, "repository": {"name": "example/repo"}, "overrides": {}}
 
 
 def profile(name):
     return load_policy(policy_root(), name)[1]
-

tests/test_evaluation.py

@@ -29,5 +29,28 @@ def test_native_capabilities_are_exposed(manifest):
     result = classify(policy_root(), profile("generic"), manifest)
     native = evaluate(policy_root(), manifest, result)["native_enforcement"]
     assert native["protected_branches"] is True
-    assert native["production_environment"]["prevent_self_review"] is True
+    production = native["production_environment"]
+    assert production["manual_human_promotion_required"] is True
+    assert production["automation_may_promote"] is False
+    assert production["distinct_human_reviewer_required"] is False
 
+
+def test_human_boundary_allows_manual_owner_authorization(manifest):
+    result = classify(policy_root(), profile("generic"), manifest)
+    obligations = evaluate(policy_root(), manifest, result)
+    boundary = obligations["human_production_boundary"]
+    assert boundary["human_authorization_required"] is True
+    assert boundary["manual_promotion_required"] is True
+    assert boundary["owner_manual_authorization_may_satisfy"] is True
+    assert boundary["automation_may_authorize"] is False
+    assert boundary["automation_may_promote"] is False
+
+
+def test_fresh_context_review_remains_required(manifest):
+    result = classify(policy_root(), profile("generic"), manifest)
+    obligations = evaluate(policy_root(), manifest, result)
+    assert "independent-review" in obligations["required_capabilities"]
+    assert "independent-review/v1" in obligations["required_evidence"]
+    review = obligations["independent_review"]
+    assert review["fresh_context_required"] is True
+    assert review["implementation_context_may_issue_verdict"] is False