Skip to main content

SRT-1 — Cognitive Operating System for Software Repositories. Anti-hallucination guardrails for AI code assistants.

Project description

SRT-1 CORE — Open Source Edition

Brain Over the Repo — Anti-hallucination, architectural coherence, and cryptographic guardrails for AI-assisted software work.

Python 3.9+ License: Apache 2.0 PyPI version


What Is SRT-1?

SRT-1 solves the problem of AI code assistants hallucinating architecture or breaking dependencies. It locally scans your codebase, builds a deep Abstract Syntax Tree (AST) manifest, and uses that context to forcefully guide the AI.

The srt1-core package provides the Memory Minimum Architecture—a completely decoupled, local-first engine that runs entirely on your machine via SQLite, without pinging a cloud API.

The "Drop-In" Sandbox (Zero Global State)

Unlike massive IDE extensions or global CLI tools that muddy your entire system, SRT-1 is hermetically sealed. You simply drop the SRT1-CORE-OSS folder into any codebase, and tell your existing AI assistant (Cursor, Copilot, etc.) to run it.

  • Auto-Isolated Ports: The engine dynamically finds available network ports, meaning you can drop SRT-1 into 5 different projects simultaneously without a single collision.
  • Local Context Only: The engine builds its .srt1 memory cache purely inside the project boundary. Zero cross-contamination across your machine.
  • The AI Handcuffs: You aren't replacing your AI; you are dropping a constraint engine into your folder. Tell Cursor to use SRT-1, and the AI is instantly handcuffed to your deterministic, architectural rules.

The Cryptographic Trust Layer

Every action in the SRT-1 pipeline is cryptographically tied to the codebase using Seed Signature. From the moment the AST generates the code manifest, to the exact moment you approve a task dispatch, the engine signs and verifies the payload. This ensures an indisputable, immutable execution roadmap.

The Local Developer Experience

SRT-1 ships with a fully-featured local Developer Dashboard and a Mobile-reflective PWA.

The Human-in-the-Loop Pipeline

SRT-1 acts as a strict proxy between you and your code assistant (like Claude Code, Cursor, or Aider).

  1. Plant a Seed: You submit a task via the terminal or the PWA.
  2. Review Blueprint: The system generates a blueprint of the intended changes and pauses execution (auto_dispatch: false).
  3. Approve via Seed Signature: You review the blueprint in the PWA. Once you click Approve, the payload is signed via Seed Signature and dispatched to your code assistant to execute.
  4. Self-Heal: If the AI makes a mistake, the AST detects the drift, bundles the error, and automatically sends it back to the AI for self-healing.

Installation

pip install srt1-core

Quick Start

1. Index a Repository

srt1-index --repo_path /path/to/your/repo

This generates srt1_code_manifest.json—the foundational map of your codebase, verified by Seed Signature.

2. Start the Live Middleware & Dashboard

srt1-middleware --repo_path ./my_project --port 7483

This spins up the local execution engine and the Developer Dashboard/PWA.

3. Open the PWA

Navigate to http://localhost:7483/mobile.html to start planting seeds and governing your AI's execution pipeline.

The Unified Ecosystem

pip install srt1-core provides the complete Open Source ecosystem:

Component Description CLI Tools
Core Intelligence Local AST mapping, File hashing, Curation srt1-index
Pro Execution Context Bundler, Execution Engine, Self-Healing srt1-bundle, srt1-execute, srt1-heal
Platform & UI Live Middleware, Developer Dashboard, PWA srt1-middleware

Architecture

SRT-1 is deployed as a single unified wheel containing all subsystems:

srt1_code_indexer/          → Code reflection and indexing engine
srt1_pro/                   → Bundling, execution, self-heal
srt1_platform/              → Middleware, Seed Queue, bridge
developer-pwa/              → Local Developer UI & Human-in-the-Loop PWA

Requirements

  • Python 3.9+
  • Core indexing uses the Python standard library only.

License

Apache License 2.0 Author: William Darnell Jernigan IV AKA THE ORIGINAL SEED

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

srt1_core-2.2.1.tar.gz (305.5 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

srt1_core-2.2.1-py3-none-any.whl (226.6 kB view details)

Uploaded Python 3

File details

Details for the file srt1_core-2.2.1.tar.gz.

File metadata

  • Download URL: srt1_core-2.2.1.tar.gz
  • Upload date:
  • Size: 305.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.14.0

File hashes

Hashes for srt1_core-2.2.1.tar.gz
Algorithm Hash digest
SHA256 fc15ca4c9f6483fe4afb64fa7599fcf325fd8efded47be621e7883cc5c196695
MD5 ca97df9f2056789f2ac6adbc5e9fd685
BLAKE2b-256 ec769b5e3b481102dc54c9472fdb4917afa2dccf6f3187e7737e0d1584959d3c

See more details on using hashes here.

File details

Details for the file srt1_core-2.2.1-py3-none-any.whl.

File metadata

  • Download URL: srt1_core-2.2.1-py3-none-any.whl
  • Upload date:
  • Size: 226.6 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.14.0

File hashes

Hashes for srt1_core-2.2.1-py3-none-any.whl
Algorithm Hash digest
SHA256 d6d1a94fcb0895a4cb89290bdab089741177941161e3638c15357006329e0262
MD5 bae4095051906f6aae9f3eda543bf907
BLAKE2b-256 ec94466774debf3b61c938f70aa98cef650bbd1f13aaa25531c4033ef14439dc

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page