autonomousguy

AI skills for embedded automotive engineers — AUTOSAR, MISRA, ISO 26262, ECU development.

autonomousguy is a library of 30 domain-accurate AI prompt files (skills) covering the full ECU software lifecycle. Use it with Claude Code, GitHub Copilot, Cursor, Gemini CLI, or any AI tool to get consistent, expert-level assistance on AUTOSAR integration, MISRA C:2025 compliance, ISO 26262 safety analysis, requirements engineering, testing, documentation, toolchain setup, debugging, and change management.

Install once. Works with any AI tool. No lock-in.

Installation

npx (recommended)

npx autonomousguy init

Shell script — Linux / macOS

curl -fsSL https://raw.githubusercontent.com/ptsilivis/autonomousguy/master/install.sh | bash

PowerShell — Windows

irm https://raw.githubusercontent.com/ptsilivis/autonomousguy/master/install.ps1 | iex

The installer asks three questions: scope (local project or global), tool(s) (see supported tools below), and skills (all 30 or select by category). Skills land at <base>/<tool-dir>/skills/autonomousguy/<category>/<skill>.md (or .claude/commands/autonomousguy/... for Claude Code, so skills are invokable as /autonomousguy:<skill-name>).

Skill catalogue

Start with workspace/codebase-analysis on any new project — it writes .autonomousguy/CODEBASE_MAP.md, which every other skill references.

workspace — Onboarding

| Skill | What it does | |---|---| | codebase-analysis | Scans the repo; maps SWCs, BSW usage, signal flows, and ASIL zones into CODEBASE_MAP.md. |

autosar — AUTOSAR Classic

| Skill | What it does | |---|---| | autosar-integration | Aligns SWC ports with ARXML, verifies RTE API consistency, resolves integration-time errors. | | swc-development | Develops a new SWC from a port interface spec: runnable structure, RTE API calls, ISR-safe patterns. | | bsw-configuration | Configures BSW modules (Com, Dem, NvM, Dcm, MemIf, Fee) from functional requirements. | | arxml-debugging | Diagnoses ARXML schema violations, missing references, and toolchain import errors. | | rte-generation-troubleshooting | Resolves RTE generator failures — unresolved ports, conflicting timing, mode-manager conflicts. | | com-stack-debugging | Traces a missing or wrong CAN signal through PduR, Com, CanIf, and CanDrv to its root cause. |

code-quality — MISRA & Code Review

| Skill | What it does | |---|---| | misra-review | MISRA C:2025 audit: violations by rule ID, severity, and location, with compliant rewrites. | | misra-driven-development | Generates new code that is compliant-by-construction across memory, arithmetic, and control flow. | | code-review | Embedded C review: correctness, ISR safety, stack usage, shared-resource access, AUTOSAR guidelines. | | naming-conventions | Audits identifiers against AUTOSAR and project-specific conventions; produces a rename map. |

architecture — Design

| Skill | What it does | |---|---| | component-design | Designs an AUTOSAR SWC topology: components, port types, data flows, ASIL boundaries. | | uml-generation | Produces PlantUML or Mermaid diagrams (sequence, class, activity, state machine). | | interface-definition | Specifies port interfaces with data element types, init values, and an ARXML sketch. |

requirements — Requirements Engineering

| Skill | What it does | |---|---| | elicitation | Transforms an informal brief into EARS-notation requirements with acceptance criteria. | | refinement | Audits requirements for ambiguity, incompleteness, and contradictions; rewrites defective items. | | traceability | Builds a bidirectional requirements-to-implementation traceability matrix. |

safety — ISO 26262

| Skill | What it does | |---|---| | iso26262-asil | Conducts a HARA: rates Severity / Exposure / Controllability, derives ASIL and Safety Goals. | | safety-goals | Develops Safety Goals with FTTI, Functional Safety Requirements, Safe States, and ASIL decomposition. |

testing — Test Design

| Skill | What it does | |---|---| | unit-test-generation | Generates MC/DC-adequate test cases with a coverage matrix and CppUTest / Unity code. | | boundary-analysis | Applies BVA to embedded fixed-width types including overflow and off-by-one detection. |

documentation — Technical Writing

| Skill | What it does | |---|---| | doxygen | Generates Doxygen comment blocks for functions, structs, and modules. | | sw-design-doc | Produces an ASPICE SWE.3-compliant Software Design Document. | | changelog | Writes a structured release changelog and flags safety-relevant changes requiring re-verification. |

toolchain — Build & Interfaces

| Skill | What it does | |---|---| | cmake-conan | Generates CMake toolchain files and Conan profiles for ARM Cortex-M/R cross-compilation. | | can-dbc-analysis | Maps DBC signals to AUTOSAR COM PDUs; flags scaling, endianness, and cycle-time issues. |

debugging — Fault Investigation

| Skill | What it does | |---|---| | problem-report-analysis | Produces a root-cause investigation plan from a field problem report. | | targeted-debugging | Guides debugging of watchdog resets, HardFaults, stack overflows, and race conditions. |

change-management — Change Control

| Skill | What it does | |---|---| | change-request-analysis | Analyses scope, ASIL impact, and affected components; produces an implementation plan. | | impact-analysis | Traces the full ripple effect across source, ARXML, requirements, tests, and safety artefacts. |

How to use a skill

Paste into any AI tool: copy the full skill file (including frontmatter) into your tool's system prompt, custom instructions field, or chat window.

Auto-discovery (Claude Code, Cursor, Gemini CLI): after installation, skills live in .<tool>/skills/autonomousguy/<category>/. Invoke by name or via /skill-name syntax.

Recommended workflow:

1. Run codebase-analysis first on any new project.
2. Pick the skill that matches your task.
3. Chain skills naturally — e.g. elicitation → component-design → swc-development → iso26262-asil → unit-test-generation.

Supported AI tools

| Tool | Skills installed at | |---|---| | Claude Code | .claude/commands/autonomousguy/ — invokable as /autonomousguy:<skill> | | GitHub Copilot | .github/skills/autonomousguy/ | | Cursor | .cursor/skills/autonomousguy/ | | Gemini CLI | .gemini/skills/autonomousguy/ | | ChatGPT Codex | .agents/skills/autonomousguy/ | | OpenCode | .opencode/skills/autonomousguy/ | | JetBrains AI | .idea/skills/autonomousguy/ | | General / other | .autonomousguy/skills/ |

Skill format

Every skill file has YAML frontmatter (drives the installer picker) and five content sections (drive the AI):

---
name: Human-readable skill name
short: One-line description shown in the CLI picker
description: Two-sentence summary.
category: <category-slug>
tags: [tag1, tag2]
---

## Context
## Instructions
## Input expected
## Output format
## Example

Contributing

Open an issue before adding new categories (minimum two skills per category). For new skills within an existing category, a PR is sufficient.

New skill checklist:

• Add a .md file to skills/<category>/ — no registration needed, the CLI discovers it automatically.
• Include all five frontmatter fields and all five content sections.
• The ## Example must use realistic ECU inputs/outputs — no placeholders.

git clone https://github.com/ptsilivis/autonomousguy.git
cd autonomousguy && npm install
node bin/cli.js   # run the installer locally

Standards & licensing notice

The skills in this library reference and operate on two copyrighted standards:

• MISRA C:2025 — © The MISRA Consortium Ltd. Sold per-seat; not redistributable.
• ISO 26262:2018 (all parts) — © ISO. Sold per-part; not redistributable.

This library cites rule and clause identifiers (e.g., "MISRA C:2025 Rule 11.3", "ISO 26262-3:2018 §6.4.3.5 Table 4") and paraphrases their intent in our own words. It does not reproduce the rule text, the standards' normative wording, the rationale or amplification sections, the examples from the standards, or the full ASIL determination table. To apply these skills in a real project you (or your organisation) must hold a properly licensed copy of each standard — the skills are an operational aid, not a substitute for the standard itself.

If you spot any content in this library that appears to reproduce normative text from either standard rather than paraphrasing it, open an issue and we will rewrite it.

License

MIT © AutonomousGuy