Radar / AI security / Sandlock 0.8.8
Sandlock 0.8.8: deferred commit for agent sandboxes
The process-based Linux AI sandbox (no container, no VM) ships deferred commit: every run returns a changeset of what it touched, and writes only land in the real working directory after review via commit() or abort().
Why it matters
Solves the core agent-sandbox dilemma: let the model edit freely, review the diff, then decide. Lightest sandbox design (process-based, no privilege), now with a rewired --dry-run and static musl builds.
What you could build with it
Run every radar research subagent inside Sandlock with deferred commit, so scraped data and generated files are reviewed as a changeset before touching the workspace.
Does it hold up?
Real user base (400+ stars) with pip/cargo/binary install paths and a workshop paper backing the design; the deferred-commit API is documented with working Python examples. Too early to judge the 0.8.8 feature in production use - no public deployment writeups found yet.
Built with Sandlock 0.8.8
- Research paper: Sandlock at the Agentic OS Workshoparticle · The paper behind Sandlock: unprivileged-Linux sandboxing for agent code, TOCTOU handling, and preliminary benchmarks vs Docker on startup latency and Redis throughput.
- Sandlock 0.8.8 release post (Multikernel)article · Multikernel's announcement of deferred commit: the DEFER branch action holds copy-on-write writes after the process exits so code can review the diff, then call commit() or abort().
- Deferred-commit API walkthrough (Python README)github · Working Python examples of BranchAction.DEFER plus the new MCP integration for sandboxed per-call tool execution.
Learn more
First spotted on github: source.
More AI security
Cloudflare security-audit-skill: multi-phase security audits for coding agentsA coding-agent skill that runs multi-phase security audits with independently verified, machine-checkable findings.security · JEV 0.63ClawSecure: free security scanner for OpenClaw AI agent skillsFree scanner that audits OpenClaw agent skills for vulnerabilities; the maker's audit of 2,890+ OpenClaw skills found…security · JEV 0.5Google DeepMind SynthID BioWatermarking technology that embeds an imperceptible, verifiable signature into AI-designed protein sequences and…security · JEV 0.49GitHub Security Lab Taskflow Agent: autonomous LLM fuzzing for C/C++Open-source agent that automates the full fuzzing lifecycle - entry-point discovery, harness generation, AFL++ runs…security · JEV 0.47LLM Agents Can Easily Tamper With Their Own TracesAn empirical study (arXiv 2026-09-24) showing that all tested local coding-agent harnesses except Muse Code allowed…security · JEV 0.4Basin (Submersion AI)Specialized cybersecurity reasoning model launched 24 Sep 2026; top-10 on CyberGym at 80.8% (8th globally), beating…security · JEV 0.36
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.