Boris Cherny's viral tweet puts TLA+ and formal methods in the agent-coding spotlight
fhuszar · x · 2026-09-26
Context: a viral tweet (1M views) from Claude Code author Boris Cherny — who used Opus 5.5 to model parts of the Claude Agent SDK in TLA+ and Lean — brought the 30-year-old formal modeling toolkit back into the spotlight, alongside Datadog's harness-first agents post.
Reasonable's long-form post offers a practical introduction: TLA+ compactly specifies what a system may do and what must always/Eventually hold, but it verifies a model of the software, not the implementation, and its main model checker only explores finite instances. The authors argue TLA+ is just the starting point — the bigger picture is combining temporal specifications, modern proof systems, and AI agents, from modeling system behaviour to machine-checked proofs, and ultimately toward software that is specified, implemented, and verified in one loop. They also preview their own work at Reasonable toward this goal.
Related event: Claude Team's Viral Tweet Revives Interest in TLA+ Formal Verification(2 posts)→
More from coding & agent
- Repurpose your old M1/M2 Mac as a local test server with this open-source runner script — jasonkneen · 2026-09-26
- Harness engineering: the key discipline for building reliable AI agents — techNmak · 2026-09-26
- Feeding PTX ISA to LLMs: Converting Figures to ASCII Art with Gemini — bingxu_ · 2026-09-26
- Merge Claims Its MCP Connectors Match or Beat Public MCPs, Cut Tokens 24% — shensi · 2026-09-26
- Define Dimensions First: Hamel Husain's Method for Synthetic Eval Data — HamelHusain · 2026-09-26
- starrocks-mcp: read-only MCP server to query StarRocks databases via Claude — modelcontextprotocol · 2026-09-26