Show HN: Skillscript – A declarative, sandboxed language for tool orchestration

📝 Discussion Summary (Click to expand)

Agent sandbox risk – "Turing completeness becomes a liability...the blast radius of a buggy agent-authored script is the whole host." – DonHopkins
Bad templating design leads to technical debt – "PHP was a templating language...ending up with a terrible hacky poorly defined incoherent templating language that was useless to programmers." – DonHopkins
Hamstringing LLMs won’t solve the problem – "Hamstringing them isn't going to help." – DonHopkins

🚀 Project Ideas

Generating project ideas…

Provide a sandboxed runtime that confines LLM‑generated scripts to a safe execution environment, blocking arbitrary subprocesses, network, and file writes.
Mitigates the Turing‑completeness risks highlighted in the HN discussion.

Key	Value
Target Audience	Developers building AI agents that generate code
Core Feature	Automated policy enforcement with resource quotas and sandbox isolation
Tech Stack	Python, FastAPI, Kubernetes, Falco, Docker
Difficulty	Medium
Monetization	Revenue-ready: SaaS subscription per compute hour

Directly addresses the “blast radius of a buggy agent‑authored script” concern quoted by HN commenters.
Practical utility for securing CI/CD pipelines and production services.

Offer a minimal, type‑checked templating DSL that prevents dangerous constructs while remaining expressive enough for common use cases.
Solves the “Greenspun’s tenth rule” bloat by limiting feature set to a safe subset.

Key	Value
Target Audience	Product engineers creating email templates or UI fragments via LLMs
Core Feature	Declarative UI/snippet components rendered in a sandboxed WASM engine
Tech Stack	TypeScript, React, WebAssembly sandbox, Vite
Difficulty	Low
Monetization	Hobby

Echoes the HN critique of over‑engineered templating languages that eventually become insecure.
Developers would appreciate a cleaner, safer alternative for templating tasks.

Auto‑scan CI pipelines for disallowed patterns (subprocess, open ports, file system writes) in agent‑generated scripts.
Provides immediate feedback and a CI gate to block unsafe code before deployment.

Key	Value
Target Audience	DevOps teams integrating AI agents into automation pipelines
Core Feature	Rule engine with customizable policy library and real‑time linting
Tech Stack	GitHub Actions, Go, Docker, PostgreSQL
Difficulty	Low
Monetization	Revenue-ready: Free tier with paid enterprise plan

Directly tackles the security concerns raised in the discussion, offering concrete utility.
Sparks conversation around best practices for AI‑generated automation.