OpenAPPA is a deterministic guardrail engine for AI agents that tracks data flow instead of matching patterns, claiming full resistance to prompt-injection data exfiltration. It runs outside the agent's prompt and execution loop as a MIT-licensed, vendor-agnostic component, driven by one declarative config describing data sources, audiences, trust levels and authorities. Every trajectory carries a security label that only gets more restrictive, and the engine derives decisions algebraically. Backed by a NeurIPS-accepted paper.
Be the first to rate OpenAPPA. Your feedback helps other users decide.
Keep an untrusted read isolated in a disposable subagent so the parent stays clean
Let audience and trust labels propagate and tighten automatically across tool calls
Verify the whole tool graph is covered by the declarative config in CI
Similar tools in the same category — compare side-by-side
ai-rete-RAG pairs a Rete rule engine with retrieval-augmented generation so a decision stays determi...
Scry is a programmable search MCP server that lets an agent run one bounded query across 108 billion...
SolidCraft is an open-source parametric 3D CAD application written in pure Rust, available on deskto...
Guides and comparisons about code tools
Agents need guardrails. Blocklists and judge models fail because they cannot see data flow. Deterministic engines track labels across tool calls instead.
A roundup of 18 AI tools launched between August and October 2026 that reached Hacker News front page: coding agents, guardrails, on-device models, dictation.
Ten Show HN projects for running several agents at once: orchestrators, subagents in isolated worktrees, shared memory trees and browser relays.
Eight Show HN projects that handle the unglamorous layer: consumer-GPU serving, remote tensors, vector-ready storage, semantic layers and agent payments.
Which AI coding assistant actually saves you time? GitHub Copilot vs Cursor vs Codium in 2026, with real comparison and recommendations by use case.
