Using OCPP.md with AI Coding Agents

This repository gives your AI coding assistant structured OCPP 2.0.1 knowledge during development sessions. It works as a Claude Code plugin or as a standalone reference for any AI agent that can read local files.

#Installation

#Claude Code Plugin (Recommended)

Add the plugin from GitHub:

/plugin marketplace add https://github.com/alexeimoisseev/ocpp.md
/plugin install ocpp@ocpp

The plugin activates automatically when the agent detects OCPP-related code or conversations. You can also invoke it explicitly:

/ocpp                    # General OCPP assistance
/ocpp smart-charging     # Smart charging deep-dive
/ocpp transactions       # Transaction handling
/ocpp authorize          # Authorization flow

#Other AI Coding Agents

The documentation in this repository works with any AI coding agent that can read local files — Cursor, Windsurf, Copilot, Aider, or a plain LLM chat with file access.

Clone this repository: git clone https://github.com/alexeimoisseev/ocpp.md.git
Point your agent at the docs/ directory. How you do this depends on the tool:
Claude Code (without plugin): Add to your project's CLAUDE.md: For OCPP protocol reference, read files from: /path/to/ocpp.md/docs/
Cursor / Windsurf: Add the docs/ directory to your project's context or rules file
Other agents: Include the path in your system prompt or project instructions

The key files are docs/OCPP-2.0.1.md (overview + all messages) and the subdirectories under docs/ for schemas, sequences, and smart charging detail.

#What the Agent Gets

Always in context:

All 64 OCPP 2.0.1 messages listed by functional block with directions
Key data types and their purpose
The escalation model (SPEC-SILENT, VENDOR-DEPENDENT, POLICY-DEPENDENT)
A routing table to find detailed docs by topic

Loaded on demand:

Field-level schemas for all messages (types, required/optional, constraints, enums)
Multi-message sequence diagrams (boot, authorization, transaction lifecycle)
Smart charging deep-dive (composite schedules, AC/DC, ISO 15118)
Worked examples with realistic JSON payloads

#Configuration

#Escalation Strictness

The OCPP specification has areas that are silent, vendor-dependent, or policy-dependent. The plugin controls how the agent handles these.

By default, the agent uses strict mode — it stops and asks you before making assumptions. To change this, add a line to your project's CLAUDE.md:

For OCPP: use pragmatic escalation mode.

Mode	Behavior
`strict` (default)	Agent stops and asks you before making assumptions in ambiguous areas
`pragmatic`	Agent flags the ambiguity with a code comment but picks a reasonable default

Recommendation: Use strict for production/certification work. Use pragmatic for prototyping.

#How It Works

The plugin includes:

Inline skill content — A compact OCPP reference always available to the agent. Includes the full message catalog, key types, and behavioral instructions.
Bundled documentation — The complete OCPP.md reference files. The agent reads these on demand when it needs field-level detail, sequence diagrams, or worked examples.
Escalation markers — The documentation explicitly flags areas where the OCPP spec is silent or behavior is vendor/policy-dependent. The agent respects these flags per your configured strictness.

#Documentation Trust Model

Not all content in the reference docs has the same confidence level:

Tier	Source	Confidence
Schema-derived	Field names, types, enums, constraints from OCA JSON schemas	High — mechanically extracted
Schema-described	Behavioral rules from OCA schema description fields	High — direct from OCA text
Spec-knowledge	Rules from OCPP 2.0.1 Part 2, known to the AI from training data	Medium — verify against official spec
Interpretation	Worked examples, pitfall lists, sequence diagrams	Lower — treat as guidance

See Methodology for full details on how the documentation was produced.

#Supported OCPP Versions

Currently: OCPP 2.0.1

OCPP 1.6 support is planned for a future release.