Advanced Local-Governance Criteria

Compatibility note: this file keeps the historical docs/v6-scope.md path and V6-linked internal identifiers for scripts, tools, and tests. Public product language should describe this as an advanced local-governance maturity layer, not as a product version.

This criteria layer remains local-first. It excludes hosted service work, database/storage work, GitHub/PR adapters, CLI UX, and typed SDKs.

It should make architect-mcp more adaptive, auditable, and useful across longer agent workflows while staying inside the existing MCP/domain/test architecture.

Local-Governance Direction

This layer focuses on local governance maturity:

• Policy composition.
• Standards lifecycle.
• Multi-turn agent continuity.
• Review explainability at scale.
• Local report artifacts.
• Regression intelligence.
• Cross-repo pattern portability without external integrations.

Policy Composition

Goal: let clients assemble reusable policy sets without persistence or hosted policy management.

Deliverables:

• Policy bundle definitions as versioned local JSON files.
• Bundle validation, conflict checks, and coverage reports.
• Bundle preview against supplied file summaries.
• Recommended bundle selection from project brief and stack signals.
• Tests for strict, migration, frontend-heavy, backend-heavy, security-focused, and novice-friendly bundles.

Standards Lifecycle

Goal: make standards evolve safely over time.

Deliverables:

• Policy and stack-pack changelog generation.
• Rule maturity levels: experimental, recommended, strict, deprecated.
• Deprecation warnings with replacement rules.
• Compatibility checks when a contract references older pack versions.
• Report whether a new ruleset is likely to create noisy findings.

Multi-Turn Agent Continuity

Goal: improve long-running local sessions without durable storage.

Deliverables:

• Session-state summaries that clients can pass back explicitly.
• Compact continuation context for the next tool call.
• Intent-debt accumulation across supplied session summaries.
• Assumption aging within a provided session payload.
• Handoff quality scoring.

Review Explainability At Scale

Goal: keep large review outputs understandable.

Deliverables:

• Finding clustering by root cause, affected boundary, stack, and verification gap.
• Top-priority remediation paths.
• "Fix one thing first" recommendation.
• Noise-risk scoring.
• Plain-English report summaries for novice users and concise machine-readable summaries for agents.

Local Report Artifacts

Goal: generate portable artifacts without needing hosted dashboards or external systems.

Deliverables:

• Markdown review report output.
• JSON review report output optimized for CI or agent clients.
• Contract change report output.
• Standards coverage report output.
• Session handoff report output.

The MCP should return report contents; writing files remains a client decision unless a local-only tool already owns that behavior.

Regression Intelligence

Goal: make known failure modes first-class.

Deliverables:

• Catalog of known agent failure patterns.
• Mapping from failure patterns to review rules and eval fixtures.
• Regression risk score for proposed changes.
• Fixture generator suggestions for new rules.
• Coverage gaps when a rule has no positive and negative fixture.

Cross-Repo Pattern Portability

Goal: reuse lessons across repos without external memory, hosted storage, or GitHub.

Deliverables:

• Portable local pattern cards.
• Pattern-card validation and sensitivity review.
• Apply-pattern preview against a project brief or file summaries.
• Conflict checks between pattern cards and current contract.
• Token-budgeted pattern summaries.

Non-Goals

• Hosted service or hosted API implementation.
• Database/storage, including schema, persistence, migrations, review history, or memory storage.
• GitHub repository, branch, PR, issue, app, OAuth, or comment workflow.
• CLI UX.
• Typed client SDK.
• Billing, accounts, teams, dashboards, or remote policy management.

Those remain out of scope until a later backend/productization phase.