frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Show HN: AI-runtime-guard – Policy enforcement layer for MCP AI agents

https://github.com/jimmyracheta/ai-runtime-guard
2•JimmyRacheta•1h ago
I built this after realizing that AI agents with filesystem and shell access can delete files, leak credentials, or execute destructive commands — and there's no enforcement layer stopping them at the execution level.

ai-runtime-guard is an MCP server that sits between your AI agent and your system. It enforces a policy layer before any file or shell action takes effect. No retraining, no prompt engineering, no changes to your agent or workflow.

Your agent can say anything. It can only do what policy allows.

What it does: - Blocks dangerous commands (rm -rf, dd, shutdown, privilege escalation) before execution - Gates risky commands behind human approval via a web GUI - Simulates blast radius for wildcard operations before they run - Creates automatic backups before destructive actions - Full audit trail of everything the agent does

Works with Claude Desktop, Cursor, Codex, and any stdio MCP-compatible client. Default profile is basic protection out of the box — advanced tiers are opt-in.

Validated on macOS Apple Silicon. Linux expected to work, formal validation coming in v1.1.

Would love feedback from anyone running AI agents with filesystem access.

Comments

entrustai•57m ago
Interesting layer to enforce policy at. You're governing what the agent can do — filesystem, shell, execution. There's a complementary problem one layer up: governing what the agent can say before output reaches a user or downstream system.

The failure modes are different. An agent that deletes the wrong file causes immediate visible damage. An agent that outputs a guaranteed return, a clinical claim it can't support, or a sycophantic opener in a regulated context causes liability that surfaces weeks later in a compliance review.

The audit trail approach you've taken with HMAC on approvals is the right instinct for the action layer. The same logic applies to the output layer — you need to prove not just what was blocked, but that the check happened at all, against a specific versioned policy, at a specific time.

Good work on the blast radius simulation — that's the kind of deterministic pre-flight check that makes governance defensible.

Show HN: Architect-Linter – Enforce architecture rules

https://crates.io/crates/architect-linter-pro
1•sergegriimm•57s ago•0 comments

Pete Hegseth and the AI Doomsday Machine

https://robertreich.substack.com/p/pete-hegseth-and-the-ai-doomsday
1•doener•2m ago•0 comments

Show HN: RubyLLM:Agents – A Rails engine for building and monitoring LLM agents

https://github.com/adham90/ruby_llm-agents
1•adham900•2m ago•0 comments

FBI raids of LAUSD Supt.'s home and office appear tied to AI chatbot probe

https://www.latimes.com/california/story/2026-02-25/fbi-raid-lausd-search-warrants
1•cdrnsf•3m ago•0 comments

Submitle – Submit, Share, and Discover Links Online

https://www.submitle.com/
1•exchangler•4m ago•0 comments

Show HN: OpenTrace – Self-hosted observability server with 75 MCP tools

https://github.com/adham90/opentrace
1•adham900•6m ago•0 comments

AT&T Acquires CenturyLink

https://old.reddit.com/r/Portland/comments/1reucu3/this_sucks_worse_than_you_may_yet_realize/
1•fullstacking•7m ago•1 comments

Automatic Discharges of Student Loans to Proceed After Dual Court Wins

https://www.forbes.com/sites/adamminsky/2026/02/25/automatic-discharges-of-student-loans-to-proce...
2•toomuchtodo•7m ago•1 comments

Multi-agent workflows often fail

https://github.blog/ai-and-ml/generative-ai/multi-agent-workflows-often-fail-heres-how-to-enginee...
1•e2e4•9m ago•0 comments

Show HN: Open-source MCP servers for self-hosted homelab AI

1•ai_engineering•9m ago•0 comments

Show HN: PixShot – Screenshot and OG Image API

https://pixshot.dev
1•juanjosegongi•10m ago•1 comments

Lawsuit could slow Micron DRAM chipmaking project in New York

https://www.syracuse.com/micron/2026/02/whos-behind-the-lawsuit-that-could-slow-microns-chipmakin...
1•walterbell•11m ago•0 comments

Nkmc – a virtual filesystem that lets AI agents call any API with ls, cat, grep

https://nkmc.ai/
1•guoyu•12m ago•1 comments

Random Ghostty theme on each launch

https://merinids212.github.io/ghostty-random-theme/
1•merinid•12m ago•1 comments

The Factory Model: How Coding Agents Changed Software Engineering

https://addyosmani.com/blog/factory-model/
1•cdrnsf•14m ago•0 comments

The Debian PHP team includes hard coded telemetry

https://salsa.debian.org/php-team/php/-/commit/aa12fa4540c8733ab6d68763b2107f39ec48fb37
1•_RPM•14m ago•1 comments

Go-Native Durable Execution

https://www.dbos.dev/blog/how-we-built-golang-native-durable-execution
2•hmaxdml•17m ago•0 comments

Ask HN: Could you create a competitor to your company at 10% of the cost?

3•TheAlchemist•19m ago•0 comments

Five years after pay transparency law, many postings don't comply

https://www.gjsentinel.com/news/western_colorado/five-years-after-pay-transparency-law-many-posti...
2•mooreds•20m ago•0 comments

Tool can summarize a YouTube video for you

https://vydcut.com
2•gaelsk•21m ago•0 comments

Show HN: BrainDump – A daily writing prompt site

https://www.braindump.club/
1•steeferino•21m ago•1 comments

Feedback Engagement (2019)

https://infiniteundo.com/post/185224298983/feedback-engagement
1•mooreds•22m ago•0 comments

Tool use and notation as shaping LLM generalization

https://the.scapegoat.dev/tool-use-and-notation-as-generalization-shaping/
1•mooreds•23m ago•0 comments

Mummy Brown

https://en.wikipedia.org/wiki/Mummy_brown
1•linsomniac•23m ago•0 comments

Show HN: I built an LLM comment detector for HN (I got banned)

2•umairnadeem123•24m ago•0 comments

Blood Feud: Oura's Health Panels versus Whoop's Advanced Labs

https://www.wired.com/story/oura-whoop-blood-labs/
1•brandonb•25m ago•0 comments

How Long Will 50ml of Ink Last? (3 Different Nibs)

https://onepenshow.com/ink/economy
1•austinallegro•27m ago•0 comments

The Impossible Landing [video]

https://www.youtube.com/watch?v=5Nkad_6aigM
1•doener•28m ago•0 comments

Show HN: Verity – I got tired of debugging duplicate emails after job restarts

https://www.useverity.io/
1•shineDaPoker•30m ago•0 comments

Pulsar timing hints at a nearby dark matter 'sub-halo'

https://phys.org/news/2026-02-pulsar-hints-nearby-dark-halo.html
1•PaulHoule•30m ago•0 comments