By tdealer01-crypto Verified
Don't trust AI. Verify every decision. Monitor every action. Verify before execution. Audit and replay proof. Optimize costs and risk.
Run a studio-style agent control layer that turns a user goal into an explainable plan, deterministic decision flow, permission verdicts, and browser-first execution after approval. Use when the user wants to plan together first, see a separate user-facing planning pane with goals, architecture, ordered work stages, risks, and permission checkpoints, then approve the plan before live execution begins. Combines deterministic decision, a super-permission-gate, local-ops-controller rules for in-boundary authority, and browser operator behavior for real-time execution against external apps. Also covers: Deterministic Safety Gate (Z3/SMT), Replayable Governance, Compliance Evidence Pack, Runtime Governance lifecycle (Before/During/After), and Hermes Controlled Executor.
Use this skill when creating, packaging, validating, publishing, or maintaining DSG GitHub Marketplace Actions. It turns DSG control-plane, action-layer permission gates, deterministic GO/NO-GO validation, audit proof, and secure deploy checks into reusable GitHub Action products.
DSG Hybrid Agent - Multi-agent workflow executor with ROM DOM simulation + real browser fallback for DSG Control Plane
Deploy and manage DSG ONE / ProofGate control plane infrastructure on AWS using CDK. Automate the complete deployment pipeline: CloudFormation stack creation, resource validation, Secrets Manager integration, ECS cluster setup, ALB configuration, database provisioning, and post-deployment verification. Use this skill when deploying to AWS (dev, staging, production), verifying infrastructure health, documenting AWS resources, and troubleshooting CloudFormation issues. Also covers: stack naming conventions, resource tagging, environment configuration, cost tracking, and disaster recovery procedures.
**Status:** ✅ Governance-enabled
External network access
Connects to servers outside your machine
Requires secrets
Needs API keys or credentials to function
npx claudepluginhub tdealer01-crypto/tdealer01-crypto-dsg-control-plane --plugin dsg-governanceBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
The source of truth is the current. implementation in this repository.
All capability statements, architecture descriptions, feature status, security claims, compliance mappings, and operational behavior must be derived from verifiable evidence in the repository, including:
If any historical document, planning note, roadmap, draft, TODO, or archived documentation conflicts with the current implementation, the implementation takes precedence.
Deprecated documents, including historical "Do not claim yet" notes or superseded planning documents, must not be treated as the source of truth.
A capability may be documented only when it is supported by the current repository state and its corresponding evidence.
When uncertainty exists, verify against the implementation and evidence before making any claim.
💡 Don't trust AI. Verify every decision.
Monitor every action. Verify before execution. Audit and replay proof. Optimize costs and risk.
▶️ Try DSG ONE Free | 📊 View Benchmarks | Live: https://tdealer01-crypto-dsg-control-plane.vercel.app
One-Sentence Pitch: Write a policy once ("Only allow transfers under ฿50K"). DSG ONE gates every AI decision against it, proves why it decided, and exports tamper-proof evidence for audits.
The DSG Control Plane is available as a Codex plugin for Claude Code, enabling AI-native governance directly within the Claude ecosystem.
# Add the repository
/plugin marketplace add tdealer01-crypto/tdealer01-crypto-dsg-control-plane
# Install the plugin
/plugin install dsg-platform@tdealer01-crypto-dsg-control-plane
DSG Action Gate - Gate AI actions through deterministic policy evaluation
POST /api/dsg/v1/gates/evaluateDSG Compliance Evidence - Generate compliance evidence packs with audit trails
POST /api/compliance-evidence-packDSG Formal Proof - Validate policy constraints using Z3 SMT solver
POST /api/dsg/v1/proofs/prove✅ Deterministic action gating with formal constraints
✅ Compliance evidence collection (L1-L5 CCVS levels)
✅ Formal proof validation for policy constraints
✅ Immutable audit trails for all governance decisions
✅ Evidence-ready execution records for audit compliance
For complete plugin documentation, see .codex-plugin/README.md
Do not trust this README alone.
Every major capability claim must be reproducible from this repository. The implementation is the source of truth.
If evidence cannot be reproduced locally, the claim is unverified.
Run these commands to verify every major claim in this README:
# TypeScript compilation
npm run typecheck
# Linting
npm run lint
# Production build
npm run build
# Unit tests (4026 passing)
npm run test
# Coverage report (89% coverage claimed)
npm run test:coverage
# Verify deterministic module (same input → same output)
npm run verify:deterministic
# Policy verification (Z3 formal proof)
npm run verify:policy
# Local HPC verification (Z3 with Docker)
npm run verify:policy:hpc:local
# Parallel CCVS evidence pipeline (L1-L5)
npm run ccvs:hpc-parallel
# Generate CCVS evidence artifacts
npm run ccvs:emit
AGT governance hooks and MCP tools for Claude Code sessions
Runtime tool-call governance for Claude Code. PreToolUse runs a two-layer check on Bash, WebFetch, WebSearch, and MCP calls (regex deny patterns on shell/web; conformal risk scorer on MCP). PostToolUse writes a hash-chained SQLite audit record across Bash, Web, and MCP calls. Blocks and escalations pop a desktop notification (macOS/Linux). /vaara-setup configures mode, protection level, and notifications in plain language via ~/.vaara/claude-code/config.json; /vaara-stats prints a summary of the audit trail.
DashClaw governance, integration, troubleshooting, and platform intelligence for Claude Code.
Achieve flow state safely with Claude Code. Auto-approves routine work, gates risky actions, hard-blocks dangerous patterns. Dual enforcement (skill + hooks), token cap for cost governance, full audit trail. Zero dependencies.
Open-source compliance workflows powered by Probo MCP (Claude Code, Codex, OpenCode)
Runtime authorization for Claude Code — expansion-aware policy checks before every Bash, Edit, Write, and MCP tool call. Shadow mode by default.