feat: add adaptive assurance and feedback

Author: suyog19

README.md

@@ -58,6 +58,8 @@ Trusted classification derives the complete Git change set from exact `--base` a
 
 Phase 2 adds a locked [agent-execution security baseline](docs/agent-execution-security.md), a complete [GitHub reference flow](enforcement/github/reference-flow.md), executable `test-result/v2` validation basis, and structured `independent-review/v2` evidence. `explain` reports profile-aware validation categories and the execution boundary. Reference evidence travels through GitHub artifacts and out-of-band provenance rather than source commits.
 
+Phase 3 adds optional privacy-conscious outcome metrics, advisory task-delegation suitability independent of delivery profile, deterministic dependency assurance, and repository-context truth maintenance. See [adaptive assurance](docs/adaptive-assurance.md) and [delegation examples](docs/delegation-examples.md).
+
 ## Policy model
 
 - **Locked** controls cannot be changed locally (fresh-context independent review, fail-closed Protected routing, exact-revision evidence, secrets, traceability, human production boundary).

docs/adaptive-assurance.md

@@ -0,0 +1,9 @@
+# Adaptive assurance and feedback
+
+Outcome metrics are optional repository/process measurements, never employee surveillance. Contract v1 aggregates by repository, workflow, change characteristics, and period; retains events for 180 days by default; distinguishes local coding time from issue-to-merge delivery; and covers flow, quality, stability, review burden, cost, complexity, and agent-effectiveness. Missing data stays explicit. Before/after reports interpret lower end-to-end time without worse quality as benefit, local-only speed as neutral, and downstream review/CI/correction growth as bottleneck displacement. Individual ranking and lines-of-code productivity are prohibited.
+
+Delegation is advisory and independent of Lean/Standard/Protected. Clear, bounded, reversible work with deterministic tests may be autonomous; novelty, tacit knowledge, negotiation, blast radius, weak validation, or unknown high-impact inputs move guidance monotonically toward bounded assistance or human leadership. Humans may choose a more conservative level. The output records rationale, boundaries, checkpoints, and uncertainty; it grants no merge/production authority.
+
+Dependency assurance detects manifests, lockfiles, VCS sources, mutable Actions, container bases, and registry configuration. Mechanical lock refresh is Standard, while unpinned/mutable trust-boundary changes are Protected. Ecosystem-native validation commands establish registry existence, resolved identity/version, lock integrity, provenance, vulnerability/licence results, and justification—for example `{command: "npm ci && npm audit", category: integration}` should be implemented as a reviewed script because shell composition is disabled.
+
+Context truth metadata maintains operational sources of truth, not generic prose freshness. Existing static entries remain valid. High-criticality context declares an owner and deterministic verification; risk-based cadence is optional. Broken commands/files/links or overdue critical reviews fail with actionable diagnostics, while low-risk static documentation has no blanket expiry.

docs/delegation-examples.md

@@ -0,0 +1,7 @@
+# Delegation examples
+
+- Familiar, bounded, reversible test fix with strong deterministic validation: `delegate_autonomous` with readiness checkpoint.
+- Unfamiliar subsystem or architectural novelty: `assist` with human design checkpoint.
+- Ambiguous Lean redesign requiring stakeholder negotiation: `human_led`; Lean does not imply autonomy.
+- Well-specified sensitive migration with strong tests and reversible rollout: `delegate_bounded` under Protected controls; Protected does not require human typing.
+- Unknown blast radius or tacit-domain need: `human_led` until uncertainty is resolved.

policy/classification/characteristics.yaml

@@ -1,6 +1,6 @@
 lean_hints: [typo, documentation_only, metadata_only, mechanical_test_only, established_css_only]
-standard: [observable_behavior, public_contract, user_journey, information_architecture, architecture, cross_service, state_transition, validation_or_error, performance, retry, concurrency, dependency_change, generated_change]
-protected: [financial_consequence, payment, payment_provider_trust, authentication, authorization, iam, privilege_or_trust, secrets, sensitive_data, production_controls, irreversible_data, infrastructure, refund_or_cancellation, learner_entitlement, sensitive_admin_action]
+standard: [observable_behavior, public_contract, user_journey, information_architecture, architecture, cross_service, state_transition, validation_or_error, performance, retry, concurrency, dependency_change, mechanical_lock_refresh, generated_change]
+protected: [financial_consequence, payment, payment_provider_trust, authentication, authorization, iam, privilege_or_trust, secrets, sensitive_data, production_controls, irreversible_data, infrastructure, refund_or_cancellation, learner_entitlement, sensitive_admin_action, dependency_trust_boundary]
 deterministic_files:
   - {pattern: "**/auth/**", characteristic: authentication}
   - {pattern: "**/iam/**", characteristic: iam}

policy/core.yaml

@@ -1,6 +1,6 @@
 schema_version: 1
 process:
-  version: 1.1.0
+  version: 1.2.0
   source: suyog19/software-engineering-process
 controls:
   traceability:

policy/evidence/predicate-types.yaml

@@ -10,3 +10,4 @@ independent-review/v2: {capability: independent-review, passing_verdicts: [pass]
 solution-sufficiency/v1: {capability: solution-sufficiency, passing_verdicts: [sufficient]}
 human-approval/v1: {capability: product-owner, passing_verdicts: [approved]}
 deployment-release/v1: {capability: ci-automation, passing_verdicts: [pass]}
+dependency-validation/v1: {capability: ci-automation, passing_verdicts: [pass]}

policy/evidence/producer-authorization.yaml

@@ -12,3 +12,4 @@ predicates:
   solution-sufficiency/v1: {producer_classes: [authorized-human, authorized-agent], minimum: {Protected: authenticated}}
   human-approval/v1: {producer_classes: [product-owner], minimum: {Protected: authenticated}}
   deployment-release/v1: {producer_classes: [trusted-ci], minimum: {Protected: trusted}}
+  dependency-validation/v1: {producer_classes: [trusted-ci], minimum: {Lean: trusted, Standard: trusted, Protected: trusted}}

policy/metrics/definitions-v1.yaml

@@ -0,0 +1,14 @@
+version: 1
+privacy:
+  optional: true
+  aggregate_by: [repository, workflow, change_characteristics, period]
+  prohibited_uses: [individual_ranking, employee_surveillance, lines_of_code_productivity]
+  retention_days: 180
+metrics:
+  flow: [issue_to_pr_seconds, implementation_seconds, end_to_end_seconds, abandonment_rate]
+  quality: [correction_cycles, post_merge_defects, substantial_rewrites]
+  stability: [ci_retries, reverts, failed_deployments]
+  review_burden: [review_seconds, review_cycles]
+  cost: [compute_seconds, token_cost]
+  complexity: [complexity_delta]
+  agent_effectiveness: [recommendation_outcome_alignment]

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "software-engineering-process"
-version = "1.1.0"
+version = "1.2.0"
 description = "Technology-neutral engineering policy and assurance platform"
 readme = "README.md"
 requires-python = ">=3.11"

schemas/delegation-input-v1.schema.json

@@ -0,0 +1 @@
+{"$schema":"https://json-schema.org/draft/2020-12/schema","type":"object","additionalProperties":false,"properties":{"acceptance_clarity":{"type":["string","null"]},"boundedness":{"type":["string","null"]},"deterministic_validation":{"type":["string","null"]},"repository_test_strength":{"type":["string","null"]},"architectural_novelty":{"type":["string","null"]},"tacit_domain_knowledge":{"type":["string","null"]},"stakeholder_negotiation":{"type":["string","null"]},"blast_radius":{"type":["string","null"]},"reversibility":{"type":["string","null"]}}}