From 471e1a0d7ad99b6400c51d43d3ad50ec697bc59b Mon Sep 17 00:00:00 2001 From: Florian BRUNIAUX Date: Mon, 12 Jan 2026 13:10:16 +0100 Subject: [PATCH] docs: add LLM Reference section and Learning Mode documentation - Add LLM Reference section in README with curl one-liner for AI assistants - Add learning mode flags and deep_dive refs to YAML index - Add comprehensive Learning Mode documentation to guide (installation, usage, integration) Co-Authored-By: Claude Opus 4.5 --- README.md | 13 +++ claude-code-reference.yaml | 6 ++ english-ultimate-claude-code-guide.md | 136 ++++++++++++++++++++++++++ 3 files changed, 155 insertions(+) diff --git a/README.md b/README.md index 358f114..e2569d0 100644 --- a/README.md +++ b/README.md @@ -64,6 +64,7 @@ claude | Need to configure a team setup | [Power User Path](#-by-role-tailored-learning-paths) | 2h | | Need to evaluate/approve adoption | [PM Path](#-by-role-tailored-learning-paths) | 20 min | | Want to check your current setup | [Audit Your Setup](#-audit-your-setup) | 2 sec | +| Want AI assistants to know Claude Code | [LLM Reference](#-llm-reference) | 1 curl | ### ⚑ Audit Your Setup @@ -77,6 +78,18 @@ curl -sL https://raw.githubusercontent.com/FlorianBruniaux/claude-code-ultimate- **Want deeper analysis?** Use [`claude-setup-audit-prompt.md`](./claude-setup-audit-prompt.md) for personalized recommendations (~3 min). +### πŸ€– LLM Reference + +Give any AI assistant instant Claude Code expertise (~2K tokens): + +```bash +curl -sL https://raw.githubusercontent.com/FlorianBruniaux/claude-code-ultimate-guide/main/claude-code-reference.yaml +``` + +**Use cases**: Paste output into ChatGPT/Claude/Gemini, add to system prompts, or reference in Claude Code with `@claude-code-reference.yaml` + +**What's inside**: Decision trees, command reference, context zones, MCP servers, agent templates, troubleshootingβ€”optimized for machine consumption. Points to line numbers in the [full guide](./english-ultimate-claude-code-guide.md) for deep dives. + ### 🎯 By Role (Tailored Learning Paths) diff --git a/claude-code-reference.yaml b/claude-code-reference.yaml index 95c1bfc..2c6be40 100644 --- a/claude-code-reference.yaml +++ b/claude-code-reference.yaml @@ -51,6 +51,8 @@ deep_dive: troubleshooting: 7892 cheatsheet: 8265 daily_workflow: 8341 + superclaude_modes: 8634 + learning_mode: 8645 # ════════════════════════════════════════════════════════════════ # DECISION TREE (most important - en premier) @@ -62,6 +64,7 @@ decide: repeating: "create agent or command" need_docs: "Context7 MCP" deep_debug: "--think or --ultrathink" + learning: "--learn (want to understand decisions)" # ════════════════════════════════════════════════════════════════ # PROMPTING FORMULA (see deep_dive.xml_prompting for details) @@ -128,6 +131,9 @@ cli: "--dangerously-skip-permissions": "auto-accept ALL (danger)" "--debug": "verbose output" "--mcp-debug": "debug MCP servers" + "--learn": "enable learning mode (explanations)" + "--learn focus:X": "learn only for domain (git/arch/sec)" + "--no-learn": "suppress learning offers" # ════════════════════════════════════════════════════════════════ # CONTEXT MANAGEMENT - see deep_dive.context_management diff --git a/english-ultimate-claude-code-guide.md b/english-ultimate-claude-code-guide.md index 26ed635..a4fbeb1 100644 --- a/english-ultimate-claude-code-guide.md +++ b/english-ultimate-claude-code-guide.md @@ -8629,6 +8629,142 @@ SuperClaude transforms Claude Code into a structured development platform throug - Smart personas for different contexts - MCP server integration - Task management and session persistence +- **Behavioral modes** for optimized workflows + +#### SuperClaude Behavioral Modes + +SuperClaude includes configurable behavioral modes stored in `~/.claude/MODE_*.md` files: + +| Mode | Purpose | Activation | +|------|---------|------------| +| **Orchestration** | Smart tool selection, parallel execution | Auto (multi-tool ops, >75% context) | +| **Task Management** | Hierarchical task tracking with memory | Auto (>3 steps, >2 directories) | +| **Token Efficiency** | Symbol-enhanced compression (30-50% reduction) | Auto (>75% context) or `--uc` | +| **Learning** | Just-in-time skill development | `--learn` flag or "why/how" questions | + +#### Learning Mode: Installation & Usage + +Learning Mode provides contextual explanations when techniques are first used, without overwhelming you with repeated explanations. + +**Installation**: + +1. Create the mode file: +```bash +# Create MODE_Learning.md in your global Claude config +touch ~/.claude/MODE_Learning.md +``` + +2. Add the content (or copy from SuperClaude framework): +```markdown +# Learning Mode + +**Purpose**: Just-in-time skill development with contextual explanations when techniques are first used + +## Activation Triggers +- Manual flag: `--learn`, `--learn focus:[domain]` +- User explicitly asks "why?" or "how?" about an action +- First occurrence of advanced technique in session + +## Default Behavior +**OFF by default** - Activates via triggers above or explicit `--learn` flag + +When active, tracks techniques explained this session to avoid repetition. +``` + +3. Register in `~/.claude/CLAUDE.md`: +```markdown +# Behavioral Modes +@MODE_Learning.md +``` + +4. Add flags to `~/.claude/FLAGS.md`: +```markdown +**--learn** +- Trigger: User requests learning mode, beginner signals, "why/how" questions +- Behavior: Enable just-in-time explanations with first-occurrence tracking + +**--no-learn** +- Trigger: User wants pure execution without educational offers +- Behavior: Suppress all learning mode offers +``` + +**Usage**: + +```bash +# Activate for entire session +claude --learn + +# Focus on specific domain +claude --learn focus:git +claude --learn focus:architecture +claude --learn focus:security + +# Batch explanations at end +claude --learn batch +``` + +**Offer Format**: + +When Learning Mode is active, Claude offers explanations after technical actions: + +``` +git rebase -i HEAD~3 +-> Explain: rebase vs merge? (y/detail/skip) +``` + +Response options: +- `y` β†’ Surface explanation (20-50 tokens) +- `detail` β†’ Medium depth (100-200 tokens) +- `skip` β†’ Continue without explanation + +**With Token Efficiency Mode** (compressed format): +``` +git rebase -i HEAD~3 +-> ?rebase +``` + +**Integration with Other Modes**: + +| Combined With | Behavior | +|---------------|----------| +| Token Efficiency (`--uc`) | Compressed offer format: `-> ?[concept]` | +| Task Management | Batch explanations at phase completion | +| Brutal Advisor | Brutal on diagnosis, pedagogical on explanation | + +**Priority Rules**: +``` +--no-learn > --uc > --learn +Token Efficiency constraints > Learning verbosity +Task flow > Individual explanations +``` + +**Example Session**: + +```bash +$ claude --learn + +You: Refactor the authentication module + +Claude: [Reads files, implements changes] +git rebase -i HEAD~3 +-> Explain: rebase vs merge? (y/detail/skip) + +You: y + +Claude: Rebase rewrites history linearly; merge preserves branches. +Use rebase for clean history before push, merge for shared branches. + +[Continues work - won't ask about rebase again this session] +``` + +**When to Use Learning Mode**: + +| Use `--learn` | Use `--no-learn` | +|---------------|------------------| +| New to a technology | Expert in the domain | +| Onboarding to project | Time-critical tasks | +| Want to understand decisions | Already know the patterns | +| Mentoring yourself | High context pressure | #### Learning Sites