Entity trail

sondera-coding-agent-hooks

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-04-03 / github-pulse-researchersondera-ai/sondera-coding-agent-hooks: Cedar Policy Language as Cross-Agent Reference Monitor — 168 StarsFrom the 'Hooking Coding Agents with Cedar Policy Language' talk at Unprompted 2026, Sondera provides Rust hook binaries and Cedar policies that intercept every shell command, file operation, and web request across Claude Code, Cursor, Copilot, and Gemini CLI. Key innovation: normalizes different agent APIs into common event types so policies write once and apply universally. Includes YARA signature scanning, LLM-based sensitivity classifiers, and information flow control with data classification levels. 168 stars, early but architecturally novel.
2026-03-23 / projects-researchersondera-ai/sondera-coding-agent-hooks: Cedar Policy Language Security Layer for Claude Code, Cursor, Gemini CLISondera hooks intercept every shell command, file operation, and web request from coding agents and adjudicate them with the Cedar Policy Language — the same formal policy engine used in AWS. Unlike probabilistic LLM guardrails, these rules always enforce: presented at Unprompted 2026 with a live demo of blocking rm -rf on OpenClaw. Rust binaries, stateful trajectory stores, works across Claude Code, Cursor, GitHub Copilot, and Gemini CLI.
2026-03-17 / projects-researchersondera-ai/sondera-coding-agent-hooks: Cedar Policy Language + Rust Reference Monitor Blocks Malicious Coding Agent Actions Before ExecutionSondera Coding Agent Hooks intercepts every shell command, file operation, and web request made by coding agents (Claude Code, Cursor, Copilot CLI, Gemini CLI) and evaluates them against Cedar policies before execution — acting as a formal reference monitor. Built in Rust with Amazon's Cedar policy language (making policies auditable and formally verifiable), it responds directly to the OpenClaw malicious-skills crisis and the broader pattern of coding agents shipping with insufficient sandboxing. The architecture is security-primitive-first: block at the system call layer, not the prompt layer.

Graph relationships

USES
Cedar
Sondera Coding Agent Hooks uses Cedar policy language
Source finding
SUPPORTS
Cursor
Sondera Coding Agent Hooks intercepts Cursor actions
Source finding
SUPPORTS
Claude Code
Sondera Coding Agent Hooks intercepts Claude Code actions
Source finding

Source trail

GitHub

Graph sources

entity graphfindings textkg edgeskg entities

Fetching from the wire…