AUTOSAR-AGENT-SKILLS GitHub llms.txt

Bugs that compile cleanly
and blow up at integration.

Zero-dependency tooling for AUTOSAR Classic Platform and CAN communication matrices. Every script is standard-library Python, so it runs inside an air-gapped build server, a CI container, or an agent sandbox without a network round-trip or a pip install.

Python 3.8+ 0 dependencies MIT 11 + 26 check codes 65 tests

dbc-matrix-audit

DBC communication matrix auditor — bit-field overlap, out-of-range signals, left/right asymmetry, naming drift, GBK encoding. Plus two-revision diff, live frame decoding and CSV export.

arxml-dbc-consistency

ARXML comm-stack × DBC cross-validation — CAN ID binding, DLC, bit geometry, byte order, init values, PduR routing, TX/RX direction. 26 check codes, exits 1 on error so it drops into CI.

Install

git clone https://github.com/autosar-skills/autosar-agent-skills.git
cd autosar-agent-skills
python -m unittest discover -s tests      # 65 tests, standard library only

No pip install required. Every skill directory is self-contained — put skills/<name>/ on an agent's skill path and it loads, or run the script directly as a CLI tool.

Why not an existing library

cantools is a good library, but parsing throws away exactly what an audit needs: cycle times, value tables, comments and frame/signal attributes. It also silently tolerates GBK byte sequences — parsing appears to succeed while every comment is lost. These tools keep all of it and report the encoding problem explicitly.

Scope is deliberately static analysis only. No bus simulation, no runtime behaviour verification — a clean run means the configuration is internally consistent and consistent with the matrix, not that the network works.

Background

Design rationale: Designing an engineering-grade skill for an AI agent. The debugging story behind the ARXML checker: 105 errors, 99 of them were not real.