akios and Aegis
These are **complements**: Akios provides low-level runtime isolation and data protection (kernel sandboxing, PII redaction), while Aegis adds high-level policy enforcement and governance (approval workflows, audit trails, kill switches)—together they create defense-in-depth for agent authorization.
About akios
akios-ai/akios
Secure runtime for multi-agent AI. Kernel sandboxing (seccomp-bpf), real-time PII redaction, Merkle audit trails.
Implements five serialized security layers (policy validation, seccomp-bpf sandboxing, real-time PII redaction across 44 patterns, token/cost budgeting, and SHA-256 Merkle audit chains) that process every workflow step before infrastructure access. Orchestrates multi-agent workflows via YAML, with native support for LLM providers (OpenAI, Anthropic, Bedrock, Ollama), filesystem, HTTP, and database agents—all subject to allowlist policies and cryptographically logged for compliance.
About Aegis
Justin0504/Aegis
Runtime policy enforcement for AI agents. Cryptographic audit trail, human-in-the-loop approvals, kill switch. Zero code changes.
Intercepts tool calls at the gateway level using SDK auto-instrumentation, HTTP proxying, or MCP integration, classifying them against zero-config detectors (SQL keywords, path traversal, prompt injection patterns) before execution. Stores tamper-evident records via SHA-256 hash chaining with optional cryptographic signing, while offering human-in-the-loop approval workflows where high-risk calls pause and route to a web dashboard for manual decision-making.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work