# itigges22/ATLAS

Adaptive Test-time Learning and Autonomous Specialization

Repository: https://github.com/itigges22/ATLAS
Canonical: https://ross.abutalabs.com/products/itigges22-atlas
Language: Python
License: AGPL-3.0
License Family: copyleft
Topics: ai-agent, ai-harness, coding-agent, developer-tools, local-first
Last push: 2026-08-24T05:42:21+00:00

## Health v2 (maintenance only)
Score: 76/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 80, longevity 15
- inputs: {"age_days": 213, "days_push": 9, "days_rel": 58, "gap_med": 36, "n_releases_24m": 4}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2078, forks 186 (observed 2026-08-28T04:06:11.349503+00:00)

## What it is
ATLAS is a local-first coding agent that wraps compact open models with planning, candidate generation, quality scoring, sandboxed testing, and repair loops. It runs model-agnostically on your own hardware (e.g., GGUF models on NVIDIA GPUs) without hosted APIs or per-token fees.

## Use cases
- run a coding agent entirely on local hardware
- make small open models handle real software tasks
- verify generated code with sandboxed compile-test-repair loops
- avoid per-token API costs for AI-assisted coding
- adaptively allocate more compute for harder coding tasks

## When to choose
- you want a privacy-preserving, local coding agent with no cloud dependency
- you run compact open-weight models (e.g., GGUF) and want frontier-style verification around them
- you prefer AGPL-licensed tooling with sandboxed code execution

## When to avoid
- you need a hosted agent with frontier proprietary models out of the box
- you lack local GPU/hardware to run models efficiently
- you need a lightweight autocomplete assistant rather than an agentic pipeline

## Facets
- artifact type: application
- maturity: active
- function: agent-framework, llm-inference, developer-tools, testing, cli
- domain: developer-tools, large-language-models, self-hosted
- platform: python, cli, cross-platform, self-hosted
- tags: coding-agent, local-first, gguf, model-agnostic, tui, sandboxed-execution, test-time-compute, ai-agents

## Member repositories
- itigges22/ATLAS (main) score 76

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:11.349503+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T02:56:07.296019+00:00, confidence not recorded.
  - readme: https://github.com/itigges22/ATLAS (fetched 2026-08-28T04:06:11.349503+00:00, sha a6c57d8c3272)
- Data as of 2026-08-30T08:39:29.467469+00:00.