governed-execution

Tandem is the authority layer between AI agents and company tools: runtime policy for tools, memory, approvals, and audit trails.

The meta layer behind AI skills. Persistent behavioral overlays that make expected agent conduct explicit across changing tasks and capabilities.

Agent gateway for external tools — Rhumb Index scores services; Rhumb Resolve routes supported capability calls through governed execution rails.

Local-first AI agent platform with a Tauri desktop app and Rust CLI. Run governed tools, durable sessions, multi-agent workflows, modular packages, and one-shot effect permits.

Elixir SDK for the Google Antigravity CLI (agy) — stream-json headless sessions, governed launch, and stack integration via cli_subprocess_core and agent_session_manager. Standalone execute/run APIs, materialized governed credentials, multi-instance isolation, and Hex-ready docs for nshkrdotcom AI SDK consumers.

Calm VS Code coding companion — Plan/Act modes, approval-gated tools, governed parallel subagents (projection + durable receipts), MCP, BroccoliDB memory.

GAP (Governed Agentic Programming) main repository | The native instruction language of AEP (Agent Element Protocol) | Fully deterministically programmed AI agents.

Governed runtime orchestration and continuity infrastructure for consequence-bearing autonomous systems.

Deterministic execution-governance infrastructure for consequence-bearing autonomous systems.

Approve, refuse, or defer AI actions with proof.

Hands-on practical guide to runx: governed agent skills, cryptographic receipts, and skill authoring

Bridgewater's All-Weather (replicable) & Pure Alpha (a data gap). All-Weather = a low-vol risk-parity compounder the family spine already beats (corr 0.96). Governed live driver: the static All-Seasons book — but its bonds_regime overlay (right signal for its 2022 weakness) fails the bar since the book self-diversifies → OFF. [Live driver]

Listed private-equity managers (BX/KKR/Apollo/Carlyle/Ares) — a levered pro-cyclical market/credit factor (beta-SPY 1.39, credit +1.55), not diversification. Governed live driver: PE-factor book + benchmark's market_regime overlay ON — first consumer sleeve to earn it (Sharpe +0.76→+0.95, DD −45%→−31%). [Live driver]

Walkthrough + runnable runx example: make a skill's refusal as provable as its write (confidence gate, evidence binding, human approval).

Governed closed-loop robotic control that blocks stale execution and recovers under physical disturbance.

The tradable core of BRICS — the Gulf pocket (KSA/UAE/QAT) as a low-corr EM diversifier. Live driver + dollar-regime overlay (validated ON: Sharpe +0.48→+0.51, DD −35%→−20%), publishes dollar_regime.txt for the family. Paper/dry-run. [Live path]

🛡️Governed execution infrastructure with KAVACH continuity-bound authority, protected execution, and authenticated evidence.

Market internals as a risk-regime read — internals coincide (don't lead), but the composite gates SPY net of cost (Sharpe +0.87→+1.11). Regime emitter built: publishes market_regime.txt, honestly labeled vol-timing not breadth alpha. [Live path]

Approve, refuse, or defer AI actions before they run.

Elixir SDK for the Cursor Agent CLI (agent) — stream-json headless sessions, governed launch, and stack integration via cli_subprocess_core and agent_session_manager. Standalone execute/run APIs, external auth via CURSOR_API_KEY, multi-instance subprocess isolation, and Hex-ready docs for nshkrdotcom AI SDK consumers.