Genoma

GENOMA is a governance-first translational research runtime with audit-grade compliance artifacts.

Local-first AI assistant for VS Code with an engineering governance layer for safe AI-driven code changes.

Apply AI-generated changes with policy enforcement and audit trail.

• Super Chat in VS Code Chat panel
• Premium side Webview chat UI
• Local Ollama + cloud Ollama-compatible endpoints
• Multi-domain modes: Dev, Data Science, Dev+DS, DevOps/Infra, Power BI
• Policy, trust, approval, custody, and governance heatmap workflows

Language Versions

• Portuguese (Brazil): README.pt-BR.md
• Hindi: README.hi.md

Why This Extension

Origin and Purpose

GENOMA is inspired by the distributed-computing era that helped map early genomic frontiers. That experience shaped a core design principle for this runtime: scientific value depends not only on generating results, but on preserving trust in how results are produced. In practice, GENOMA focuses on governance-by-design with auditable custody, cryptographic integrity, and policy controls so translational research outputs remain reproducible, verifiable, and ethically accountable.

Most assistants stop at "generate code". This extension goes further: it helps teams govern AI code changes with policy controls, auditability, and measurable outcomes.

What's New (v4.x)

• Policy-as-code with extends, imports, and profiles (strict, balanced, experimental)
• Domain-aware risk and trust scoring
• Dual approval for high-risk overrides
• Tamper-evident chain-of-custody (HMAC + hash chain)
• Optional Git provenance (branch/commit metadata with chainId + sourceHash)
• Governance heatmap by domain in Webview
• Governance experiment metrics CLI (yarn run eval:governance)

Release Highlights (4.1.0)

• Policy Imports for reusable governance modules via imports in .ollama_policies.json
• Expanded Governance Heatmap with domain-level rollback/validation/block/override rates
• Top triggered rule per domain + clickable domain rows for governance prompt drafting

Open Super Chat

1. Open View -> Chat
2. Type @ollama-cloud-chat
3. Run /provider to select endpoint + model

Commands

Modes

• code: explain, refactor, tests, fix, review, security
• ds: eda, sql, features, train_eval, debug, doc
• devds: end-to-end software + data workflow
• devops / infra: PowerShell-first operations guidance
• pbi: pbi_model, pbi_m, pbi_dax, pbi_visuals, pbi_checks

Premium Webview Features

• Presets for common workflows
• Quick Prompts chips (one-click task starters)
• Favorites with search
• Markdown rendering (code blocks, links, lists)
• Message-level actions: Copy / Apply / Favorite / Reuse
• Retry button for last request
• Provider/model/mode/action/context controls
• Setup Rapido section: Usar Local, Conectar Cloud, Escolher Modelo Local, Testar Ollama Local
• One-click Local/Cloud switch with clearer visibility
• Cloud-selected UI shows inline Login and Set Token actions
• Local runtime health in panel (online/offline + model count)
• If Cloud is selected without token, chat auto-falls back to Local Ollama

Governance Layer

• Policy guardrails before apply (.ollama_policies.json)
• Domain-level enforcement and explainability
• Trust-based gating from observed behavior
• Override audit with justification + optional dual approval
• Tamper-evident custody trail:
  • .ollama_copilot_apply_chain.jsonl
  • chainId, signatures, prevHash, entryHash
• Optional Git provenance for repository-level evidence
• Domain heatmap for operational tuning

Key Settings (Governance)

• ollamaCopilot.policyEnabled
• ollamaCopilot.policyFile
• ollamaCopilot.policyProfile
• ollamaCopilot.policyRiskThreshold
• ollamaCopilot.policyTrustEnabled
• ollamaCopilot.policyDualApprovalEnabled
• ollamaCopilot.policyCustodySigningEnabled
• ollamaCopilot.policyGitProvenanceEnabled

Cloud + Onboarding

• First-run guided onboarding for Ollama Cloud login + API keys
• First run auto-opens Genoma sidebar after install
• Commands in Command Palette:
  • Genoma: Open Ollama Cloud Signup / Keys
  • Genoma: Open Cloud API Keys
  • Genoma: Set Cloud API Key

Reliability & Product Quality

• Action validation by mode (Webview + Super Chat)
• Structured output templates for review/devops workflows
• Legacy extension upgrade prompt (optional uninstall)
• Settings migration/versioning support
• Optional local telemetry (opt-in): .ollama_copilot_telemetry.jsonl
• Optional audit log: .ollama_copilot_log.jsonl

International Readiness

• International compliance baseline: docs/INTERNATIONAL_COMPLIANCE.md
• Data source license registry: DATA_LICENSES.md
• License metadata schema (for ingestion enforcement): tools/compliance/data-license-entry.schema.json
• Federated join governance policy: docs/FEDERATED_JOINS.md
• Federation event schema (for join audit enforcement): tools/compliance/federation-event.schema.json

Biomedical International Framework

• Designed for multi-jurisdiction biomedical projects with traceable governance workflows.
• Supports a federated data-core strategy across Ensembl, NCBI/ClinVar, gnomAD, UniProt, and Open Targets.
• Enforces per-source metadata for license, dataset version, terms URL, attribution, and redistribution.
• Provides an engineering baseline for GDPR, LGPD, and HIPAA-aligned operations.
• Keeps provenance-ready records for high-impact actions and dataset lineage.
• Includes release-gate documentation for compliance review and legal handoff.

Compliance and legal docs:

• docs/INTERNATIONAL_COMPLIANCE.md
• DATA_LICENSES.md
• tools/compliance/data-license-entry.schema.json
• docs/FEDERATED_JOINS.md
• tools/compliance/federation-event.schema.json
• docs/RESEARCH_PROPOSAL.md
• docs/EXPERIMENT_PROTOCOL.md
• docs/PROPOSTAS_MESTRADO.md
• docs/PRE_PROJETO_MESTRADO.md
• docs/LANDMARK_CRITERIA.md
• docs/GENOMA_RUNTIME.md
• docs/CITATION_VALIDATION.md

Federated Biomedical Join Governance

• Federated joins are policy-evaluated before model inference.
• Join events require auditable metadata: sources, purpose, crossBorder, combinedRiskScore, policyVerdict, approvedBy.
• Post-join inference must include custody linkage via federationEventId.
• Baseline rule set is provided in .ollama_policies.json under biomedical_federation.
• Local federation event template: .genoma_federation_event.example.json
• Pre-apply policy CLI check: yarn run eval:federation --eventFile .genoma_federation_event.json
• Functional federated runtime demo: yarn run genoma:demo
• Full research run with persisted artifacts: yarn run genoma:run
• Full research run with strict citation gate: yarn run genoma:run:strict
• Reproducible research report: yarn run genoma:report
• Benchmark suite: yarn run genoma:benchmark
• Citation DOI/PMID validation gate: yarn run eval:citations

Note: this is an engineering compliance baseline, not legal advice.

Governance Experiment

yarn run eval:governance --splitDate 2026-02-01T00:00:00Z --out governance-report.json

This produces domain-level metrics for:

• rollback rate
• validation success rate
• block/override rate
• baseline vs governed deltas

Development

yarn compile
yarn test

Support / Donations

If GENOMA helps your research, you can support continued development:

• Email: t7426541@gmail.com
• WhatsApp: +55 42 99127-4857
• Pix (donation): +55 42 99127-4857

Local-first

Supports:

• Local Ollama (http://localhost:11434)
• Remote Ollama-compatible APIs