Build a language interpreter around executable examples and sanitizers
At a glance
Ankur Sethi used Claude Code to build Cutlet, steering generated C code with written plans, example-program outputs and memory-safety checks. This workflow is listed under language-implementation, agent-feedback, memory-safety.
His report, linked from Lobsters, describes reviewing feature designs and saving small implementation plans before Claude implemented them in Docker. The loop combined failing tests, expected outputs for sample programs, clang-tidy and a make test-sanitize target using ASan and UBSan. Token, AST and bytecode dumps helped the agent locate errors.
How it went
Sethi reports a four-week project that runs real programs on macOS and Linux. The public repository includes plans, tests and examples.
Worth knowing
He describes Cutlet as a proof of concept and says he did not review the generated code. At publication it lacked file I/O and error handling.
Save ideas in this browser. Upvotes count across the board.
Show HN: Abyss runs Pi/Hermes/Claude/Codex (any ACP agent) in Docker with path-aligned mounts, then applies host WASM plugins that can redact secrets, hydrate RAG, inject MCP, or auto-cancel sessions—agent-agnostic harness features without rewriting each agent plugin.
An indie hacker packaged an agency's eight repeated workflows (briefs, client approval with revision counters, quotes with margin checks, content calendars, post-mortems) as free MIT Claude skills.
A builder asked Grok Bot to bridge to Claude Code—Grok does live research/browser work while Claude Code builds, and they ping each other for Playwright and research help.
Indie Hackers founder Sean Prentice used Claude Code to build an orchestrator + marketing director + specialist agents (content/ads/research) for his SaaS; outputs land in Slack and nothing posts or spends without his explicit say (~$10/mo self-use).