Concordat Enforces Typed Protocols for LLM Agents

// 1h agoRESEARCH PAPER

Concordat Enforces Typed Protocols for LLM Agents

Concordat is an open-source Python library that enforces finite-state protocols on LLM agents, limiting tools to legal actions and blocking unsafe calls before side effects. Its tau-bench evaluation eliminated unconfirmed airline changes while maintaining comparable task success.

// ANALYSIS

Concordat makes a compelling case that agent safety should be enforced structurally, not entrusted to prompts alone.

–Withheld tools prevent models from bypassing required confirmation steps.
–Typed protocols bring typestate, session types, and safe inheritance into agent development.
–The evaluation found zero unconfirmed changes versus 13 GPT-4.1 tasks and 29 Claude Haiku task-runs for standard agents.
–Budget splitting across handoffs adds useful control over multi-agent cost and resource usage.
–The project is promising but remains pre-alpha and tested primarily on a narrow airline workflow.

// TAGS

concordatagentllmtool-useguardrailsframeworkopen-sourceresearch

DISCOVERED

1h ago

2026-10-29

PUBLISHED

2h ago

2026-10-29

RELEVANCE

9/ 10

AUTHOR

arnaudmercier

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

UPDATE19m ago

Shopify Opens Billion-Product Catalog to Agents

Shopify’s Fall ’26 Edition lets AI agents search more than one billion products through Shop.md, complete purchases with Shop Pay, and interact with every Liquid storefront through structured WebMCP tools. Australian merchants also get early access to Shop Pay Installments.

RESEARCH1h ago

WorkFactory maps intents into executable agent tasks

WorkFactory is a framework for generating coherent professional tasks and expert trajectories for training work-oriented AI agents. Its authors built 112,686 tasks across 612 occupations and fine-tuned the WorkPilot model family, with the largest model improving substantially on professional-work benchmarks.

RESEARCH1h ago

TCM-DiagCommit Reframes TCM Diagnosis Benchmarks

TCM-DiagCommit introduces a benchmark for testing evidence-conditioned syndrome selection, staged diagnostic updates, and commitment under uncertainty. Across five zero-shot LLMs, accuracy fell from 63.2–66.4% on random candidates to 14.8–22.3% across all 148 labels, while wrong-terminal commitments reached 62.2–74.3%.