Mcp-Omega-Brain
AI agent provenance, trust, and auditability layer. VERITAS multi-gate scoring, Cortex approval gates, S.E.A.L. hash-chain audit ledger, and semantic RAG with cryptographic provenance tracking for every decision an agent makes.
README
<div align="center"> <img src="https://raw.githubusercontent.com/VrtxOmega/Gravity-Omega/master/omega_icon.png" width="100" alt="VERITAS" /> <h1>OMEGA BRAIN MCP + VERITAS BUILD GATES</h1> <p><strong>Standalone Model Context Protocol Server with Built-In Intelligence + Deterministic Build Pipeline</strong></p> <p><em>Two files. One dependency. Full sovereign cognition + 10-gate build verification.</em></p> </div>
A self-contained MCP server that gives AI agents verifiable provenance, cross-session memory, a cryptographic audit trail, and a 10-gate deterministic build evaluation pipeline. No external server required.
SYSTEM INVARIANT: VERITAS Build does not determine whether code is 'good.' VERITAS Build determines whether code survives disciplined attempts to break it under explicitly declared primitives, constraints, test regimes, boundaries, cost models, evidence, and policy.
Architecture
| Layer | Component | Role |
|---|---|---|
| Brain Core | Vault (SQLite) | Persistent session/entry storage with FTS5 search |
| Brain Core | SEAL Ledger | Append-only SHA3-256 hash chain for tamper-proof audit |
| Brain Core | RAG Provenance | Semantic embedding store with 3-tier engine (ST/fastembed/TF-IDF) |
| Brain Core | Cortex | Tri-Node approval gate with steer/block modes |
| Brain Core | Handoff | SHA-256 sealed cross-session memory transfer |
| Build Gates | 10-Gate Pipeline | INTAKE→TYPE→DEPENDENCY→EVIDENCE→MATH→COST→INCENTIVE→SECURITY→ADVERSARY→TRACE/SEAL |
| Build Gates | Evidence Engine | Quality(e) formula, MIS_GREEDY, Agreement computation |
| Build Gates | CLAEG | Constraint-locked state machine with 3 terminal states |
| Build Gates | NAFE Scanner | Narrative failure signature detection |
Verdict System
| Verdict | Precedence | Meaning |
|---|---|---|
PASS |
0 (lowest) | All gates satisfied. Artifact is deployable under declared regime. |
MODEL_BOUND |
1 | Gates pass but resource/coverage/confidence near redline. Deploy with monitoring. |
INCONCLUSIVE |
2 | Insufficient evidence or timeout. Cannot affirm or deny. Block deploy. |
VIOLATION |
3 (highest) | Constraint failure, security vulnerability, or test failure. Block deploy. |
Tools (26)
Brain Core (12)
| Tool | Purpose |
|---|---|
omega_preload_context |
Episodic task briefing: RAG + vault + sealed handoff + VERITAS score |
omega_rag_query |
Semantic search over RAG provenance store |
omega_ingest |
Add text fragment to RAG store |
omega_vault_search |
Full-text keyword search across vault entries |
omega_cortex_check |
Tri-Node approval gate with similarity scoring |
omega_cortex_steer |
Cortex correction mode — steer drifting args or hard block |
omega_seal_run |
Append tamper-proof S.E.A.L. entry to audit ledger |
omega_log_session |
Write session record to vault |
omega_write_handoff |
SHA-256 sealed cross-session handoff |
omega_execute |
Cortex-wrapped meta-tool — default execution path |
omega_brain_report |
Human-readable audit report |
omega_brain_status |
Unified brain health: vault stats, fragment count, ledger entries |
Build Gates (15)
| Tool | Purpose |
|---|---|
veritas_intake_gate |
Gate 1/10: Canonicalize, validate fields, compute ClaimID |
veritas_type_gate |
Gate 2/10: Primitives, domains, operators, symbols |
veritas_dependency_gate |
Gate 3/10: SBOM, CVE, integrity, licenses, depth |
veritas_evidence_gate |
Gate 4/10: MIS_GREEDY, Quality(e), K/A/Q thresholds |
veritas_math_gate |
Gate 5/10: Constraint satisfaction via interval arithmetic |
veritas_cost_gate |
Gate 6/10: Resource utilization vs redline thresholds |
veritas_incentive_gate |
Gate 7/10: Source dominance and vendor concentration |
veritas_security_gate |
Gate 8/10: SAST, secrets, injection, auth, crypto |
veritas_adversary_gate |
Gate 9/10: Fuzz, mutation, exploit, outage, spike |
veritas_run_pipeline |
Full 10-gate pipeline — final verdict + seal hash |
veritas_compute_quality |
Compute Quality(e) for single evidence item |
veritas_mis_greedy |
Run MIS_GREEDY algorithm on evidence items |
veritas_claeg_resolve |
Map verdict to CLAEG terminal state |
veritas_claeg_transition |
Validate state transition (absence = prohibition) |
veritas_nafe_scan |
Scan for NAFE failure signatures in AI text |
Resources (9)
| URI | Description |
|---|---|
omega://session/preload |
Auto-fetched startup: RAG + handoff + vault context |
omega://session/handoff |
SHA-256 verified cross-session handoff |
omega://session/current |
Session ID, call count, data directory |
omega://brain/status |
DB stats, embedding engine, ledger count |
veritas://spec/v1.0.0 |
Full canonical VERITAS Omega Build Spec |
veritas://claeg/grammar |
Terminal states, transitions, invariants, prohibitions |
veritas://gates/order |
The 10-gate pipeline sequence |
veritas://thresholds/baseline |
Dev/baseline regime numeric thresholds |
veritas://thresholds/production |
Escalated production regime thresholds |
Quick Start
Requirements
- Python 3.11+
pip install mcp
Optional (better embeddings)
pip install fastembed # ONNX embeddings, ~30MB
pip install sentence-transformers numpy # Best quality, larger
Configure in Claude Desktop / Antigravity
{
"mcpServers": {
"omega-brain": {
"command": "python",
"args": ["path/to/omega_brain_mcp_standalone.py"],
"env": { "PYTHONUTF8": "1" }
}
}
}
SSE Mode
python omega_brain_mcp_standalone.py --sse --port 8055
# Endpoints: GET /sse, POST /messages
File Structure
omega-brain-mcp/
omega_brain_mcp_standalone.py # MCP server (~1430 lines) — Brain Core + tool dispatch
veritas_build_gates.py # Gate engine (~1430 lines) — pure deterministic logic
omega_client.py # Python client helper
requirements.txt # mcp>=1.0.0
pyproject.toml # Package config
tests/
test_build_gates.py # Gate pipeline tests
test_veritas.py # VERITAS scoring tests
test_seal.py # SEAL chain integrity tests
test_handoff.py # Handoff seal/context tests
test_cortex.py # Cortex approval tests
test_vault.py # Vault persistence tests
CLAEG State Machine
INIT → { STABLE_CONTINUATION | ISOLATED_CONTAINMENT | TERMINAL_SHUTDOWN }
STABLE_CONTINUATION → { STABLE_CONTINUATION | ISOLATED_CONTAINMENT | TERMINAL_SHUTDOWN }
ISOLATED_CONTAINMENT → { STABLE_CONTINUATION | TERMINAL_SHUTDOWN }
TERMINAL_SHUTDOWN → {} (absorbing)
Invariant: Absence of an allowed transition is treated as prohibition.
License
MIT
<div align="center"> <sub>Built by <a href="https://github.com/VrtxOmega">RJ Lopez</a> | VERITAS Omega Framework</sub> </div>
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。