Sandboxed code execution daemon with enforced memory, CPU, and output limits. JSON API for humans and AI agents.
Security middleware for AI Agents. Intercepts shell commands before execution using a multi-layer pipeline: binary allowlist, regex patterns, deterministic intent coherence mapping, and LLM semantic check as last resort.
Safe execution layer for LLM tools: fatigue-aware gates, witness logs and policy checks before agents touch real systems.
Control plane for multi-agent engineering work: intent, plan, execute, review, synthesize, with trajectories and routing policy.
Deterministic command inspection engine and defense-in-depth safety extension for Pi coding agent.
Aisandbox — AI monolith dev sandbox
Local natural-language command router that converts AI intent into whitelisted Git and workflow actions.
Deterministic MCP runtime for safe tool execution, policy gates, contracts, and local AI workflow governance.
Terminal-native AI agent orchestrator with safety gates, repo intelligence, and controlled execution workflows.
A v1 command execution sandbox in Zig with timeout control, output capture, and process handling.
Güvenli ve kontrollü yerel masaüstü ajan sistemi (simülasyon + kullanıcı onayı) v1.4
Safe file changes for AI agents with stale-state protection, bounded filesystem authority and evidence-based outcomes.
Results utilities and wrappers made for JS
Safe execution layer for AI-agent VPS operations with bounded authority, state-bound execution and evidence-based outcomes.
Safe PR merge execution for AI agents with stale-state protection, bounded authority and evidence-based outcomes.
Safe email execution for AI agents with bounded authority, stale-state protection, same-instance duplicate suppression and evidence-based outcomes.
Run learner code safely-ish - restricted builtins, no imports, a timeout, zero dependencies
Think. Review. Code. Safely.
Default-deny local execution and model-runner boundary for broker workflows, with bounded JSON results, resource controls, model verification and JSONL streaming.
Safe-by-construction local execution substrate for mini-swe-agent-style bash-action coding agents.
To associate your repository with the safe-execution topic, visit your repo's landing page and select "manage topics."