Project STROT

Project STROT is an open decision compiler that compiles intent into type-safe Python artifacts for zero-cost record classification and routing.

Project STROT is an open decision compiler that compiles user intent once using a language model into a frozen, self-contained Python artifact. Production workloads run locally on hardware using a small decision model to arbitrate records through classification, scoring, and routing without generating new text at runtime. It handles unstructured text for tasks such as cyber incident triage and financial complaint routing. The architecture features calibrated predictions where low-confidence records abstain rather than guess, and supports model swapping with local models like Julia-1 or hosted APIs. It outputs type-safe JSON with probabilities and confidence scores.

Project STROT at a glance

Best for: Software engineer
Runs on: Python, self-hosted, open source
Pricing: Free
Released: October 2026
Category: AI coding tool
Website: github.com/strot-ai/project-strot

Key features

Compile-time intent compilation
Type-safe JSON output generation
Confidence-based record abstention
Content-addressed artifact cache
Fitted guarantees with temperature scaling
Model-independent decision model slot

Use it for

Classifying unstructured cybersecurity incident notes
Routing financial consumer complaints to products
Executing typed decisions with probabilities
Running local decision models at zero marginal cost

Alternatives to Project STROT

jevgrep — Reduces coding agent costs by providing targeted, relevant repository context.
Stof — Runs AI-generated logic safely in a sandbox inside the app.
Nouli — It gives stateless decision models a persistent memory across branching sessions.
Cairn — Solves open-ended problems using a blackboard architecture and OODA loops.
OpenCode — Works with 75+ model providers, local models, and existing subscriptions
Decisions API — Fast classification and routing decisions without generating free text.
Stroq — Scans recorded sessions and guards future actions with local hooks.
Grit — Automates large-scale code refactoring and migration tasks efficiently.
pstack — Improves AI coding agent outcomes with rigorous workflows.
Codeflash — Automated performance engineering that identifies and implements code optimizations.
Weft — It enables safe parallel coding agent work without file locks.
Sgrep — Runs locally without accounts or hosted indexes using small CPU models.

All Project STROT alternatives