#guardrails

The immune system for AI coding agents

The agent framework where the model never holds the trigger — every consequential action clears your policy first, waits for a human when it matters, and lands on a record you can verify. Build on it, or put it around the agent you already have.

Deterministic Stop-gate and code-graph context for AI coding agents: blocks only net-new findings and gives the loop a structural map of the codebase, locally, with no model in the gate.

Self-hosted PII redaction proxy for LLM APIs and agent CLIs. Scrubs names, emails and secrets from Claude Code, Codex and any OpenAI-compatible app before requests leave your machine, tool-call arguments included, then restores them in the reply. Local detection, zero telemetry.

Superagent protects your AI applications against prompt injections, data leaks, and harmful outputs. Embed safety directly into your app and prove compliance to your customers.

BonkLM - LLM Security Guardrails with Interactive Setup Wizard

Zero-trust governance layer for Claude Code. Policy-enforced MCP gateway with autonomy controls, middleware chain, audit log, HALT kill-switch, and prompt-injection defense.

Production plumbing for LLM apps: context, cost, testing & governance — composable, framework-agnostic Python libraries.

Streaming privacy for LLM apps: pii-leak-benchmark (does your gateway leak PII in streams?), chunk-invariance (one-line test for split-boundary leaks), mcp-ssrf-check, and a self-hosted redacting gateway.

Stop repeating yourself to Claude.

EgisAI SDK

The authorization kernel for AI agents. Block, allow, or escalate agent tool calls with YAML rules — deterministic-first, LLM fallback.

Validate and retry LLM outputs for ruby_llm. Describe the JSON response you expect, fall back to a stronger model when the cheaper one fails the rules, and gate CI on regressions — all as one contract object per step.

Deterministic governance for Claude and MCP tool calls. Pin approved capabilities, detect drift, and refuse unsafe or unapproved actions before execution. No model in the decision path.

Archkeel checks architecture boundaries and declared changes in AI-assisted code.

Umbrella installer and combined report over the independently installable guardrail gates

Two-layer enforcement (pre-commit hook + CI mirror) for small teams using AI agents. Catches debug leaks, file growth, secrets, and forbidden patterns before they merge, and scaffold-doctor tells you when a guardrail is installed but not actually running.

Agentic Runtime

Constraint-driven runtime for TypeScript. Declare requirements, let the runtime resolve them.