fxcodebase-backup

Installing the Indicore Reader skill

The skill lives in skills/indicore-reader/. It is a standard Agent Skill — a SKILL.md plus a references/ and assets/ folder — and needs no build step or dependencies.

Option A — Claude Code (project skill)

Copy the skill into your project’s .claude/skills/ directory:

mkdir -p .claude/skills
cp -r skills/indicore-reader .claude/skills/

It is now available in that project. Claude will consult it automatically when you hand it Indicore Lua and ask to port/convert/understand it. You can also invoke it explicitly with /indicore-reader if your setup exposes skills as slash commands.

Option B — Claude Code (personal skill, all projects)

mkdir -p ~/.claude/skills
cp -r skills/indicore-reader ~/.claude/skills/

Option C — Claude.ai / Claude Desktop (uploaded skill)

Zip the skill folder and upload it wherever your Claude client accepts custom skills:

cd skills && zip -r indicore-reader.zip indicore-reader && cd ..

Then add indicore-reader.zip through your client’s skill/upload UI.

Verifying it works

Give Claude an Indicore .lua file and ask, for example:

Port this fxcodebase indicator to a platform-neutral spec.

Claude should recognise the Indicore code, read the relevant files under skills/indicore-reader/references/, and return a filled-in Port Specification. If it doesn’t pick the skill up automatically, mention “Indicore” or “fxcodebase” explicitly, or invoke it by name.

Triggering — read this

Skills are consulted at the model’s discretion, and a strong model (Opus 4.8+) that already reads Indicore well will often decline to auto-invoke this skill because it believes it can do the task unaided (we measured ~100% precision but low auto-recall, unchanged by description wording). That’s fine — but for reliable, consistent results:

• Invoke it explicitly when you want it: /indicore-reader (or mention “use the indicore-reader skill” / “use the Indicore porting reference”). This is the recommended default for a porting run.
• It auto-triggers most reliably on harder artifacts (signals, owner-drawn/heatmap indicators, multi-timeframe studies, trading strategies) and on smaller/cheaper models — which is exactly where it adds the most value (see the “Strengths” section of the README).
• In an automated porting pipeline, make reading SKILL.md an explicit first step rather than relying on auto-invocation.

Updating

Re-copy the folder after pulling changes. Nothing is cached or compiled; the Markdown files are read directly.

Notes

• The skill is fully self-contained — it does not require the original Indicore SDK docs at runtime; everything it needs is in references/.
• No network access, API keys, or external tools are needed.

This site is open source. Improve this page.