bookmap-addon-dev

An Agent Skill that turns a natural-language trading idea into a complete, buildable Bookmap addon — a Java module written against the Bookmap Layer 1 (L1) API, packaged as a .jar, and loaded via Settings → Configure add-ons. Agent Skills are an open format supported by Claude Code and many other AI coding agents.

Describe what you want ("flag when large orders hit the book", "plot a custom moving average over the heatmap") and the agent scaffolds the project, writes the addon, builds the jar, and tells you how to load it. This repo is the source of the skill, not an addon itself.

What you can do with it

Once installed, ask your agent (Claude Code shown in the examples) to:

• Check your environment — verify JDK/Gradle and locate your Bookmap install.
• Start a new addon project — scaffold a buildable Gradle project with a working starter addon.
• Build an addon from a trading idea — indicators, alerts, markers, strategies, screen painters.
• Compile it to a jar — ready to load into Bookmap.
• Look up the API — exact class locations and method signatures for your Bookmap version.
• Review / debug existing addon code — including the "why won't my jar load" checklist.

Requirements

• JDK 17+ (with javac) — the only hard requirement to build an addon.
• Python 3.8+ — for the skill's helper scripts (standard library only, no pip install).
• Bookmap (installed) — optional for building; needed to load/run/debug and for offline builds.
• Internet, once — for the Gradle wrapper / Maven fetch; offline mode needs neither.

See SETUP.md for per-OS install commands and all installation details. The Quick start below provides the initial steps.

Quick start

From this folder:

# 1) install the skill (default: Claude Code's ~/.claude/skills/; --dest <dir> for other agents)
python3 scripts/install_skill.py

# 2) verify your environment
python3 scripts/check_prereqs.py

# 3) scaffold a new project (creates ./large-order-flagger, never touches the current dir)
python3 scripts/new_project.py --name "Large order flagger" --package com.acme.bookmap --flavor simplified

# 4) build it
python3 scripts/build_addon.py ./large-order-flagger        # or: cd large-order-flagger && ./gradlew jar

# 5) load build/libs/large-order-flagger.jar in Bookmap:
#    Settings -> Configure add-ons -> Add -> select the addon -> enable

# 6) make it YOUR addon: open your coding agent inside the project and describe what you want
cd large-order-flagger && claude     # or any skill-compatible coding agent
#    e.g. "flag when orders larger than 500 lots hit the book" — the agent writes the code;
#    then rebuild (step 4) and reload (step 5)

Steps 2–5 are what the agent runs for you once the skill is installed — after step 1, you can simply describe the addon you want, or invoke /bookmap-addon-dev, and go straight to step 6's edit → rebuild → reload loop.

How it works

The skill is a folder the agent loads on demand: when you mention a Bookmap addon (or just describe a trading idea), the agent follows the guided workflow in SKILL.md, running the bundled scripts for the mechanical parts (environment checks, scaffolding, building) and consulting the bundled API references so it never guesses signatures. The output is a normal Gradle project you own — Python is only used by the helper scripts, not by the addon at runtime.

Working on the skill

Want to improve the skill itself (its instructions, references, scripts, or project template)? See CONTRIBUTING.md.