continuous-improvement 1.0.0 → 1.1.0

package/README.md

@@ -2,174 +2,153 @@
 
 ### Install discipline into your AI agent in one command.
 
-AI agents are useful. They're also reckless by default.
+<p align="center">
+  <img src="promo/combined.gif" alt="Before vs After continuous-improvement" width="700" />
+</p>
 
-**continuous-improvement** makes an agent research first, plan before coding, verify before saying "done", and reflect so it stops repeating the same mistakes.
+AI agents are useful. They're also reckless by default — they skip docs, overbuild, claim success before testing, and repeat the same mistakes.
 
-## Install
+**One command fixes it:**
 
 ```bash
 npx continuous-improvement install
 ```
 
-If auto-detect misses your setup:
+That's it. Your agent now follows a structured loop: **Research → Plan → Execute → Verify → Reflect → Learn → Iterate.**
+
+---
+
+## Supports every major AI coding tool
+
+| Tool | Flag | What it does |
+|------|------|-------------|
+| **Claude Code** | `--claude` | Appends rules to `CLAUDE.md` + installs Mulahazah learning hooks |
+| **Codex** | `--codex` | Appends rules to `AGENTS.md` |
+| **Cursor** | `--cursor` | Appends rules to `.cursorrules` |
+| **OpenClaw** | `--openclaw` | Installs skill to `~/.openclaw/skills/` |
+| **ChatGPT** | `--chatgpt` | Prints block to paste into Custom Instructions |
+
+Auto-detection works — just run `npx continuous-improvement install` and it finds your setup. Or specify:
 
 ```bash
 npx continuous-improvement install --claude
-npx continuous-improvement install --codex
 npx continuous-improvement install --cursor
-npx continuous-improvement install --openclaw
 npx continuous-improvement install --chatgpt
 ```
 
-Optional global Claude install:
+Global Claude install (applies to all projects):
 
 ```bash
 npx continuous-improvement install --claude --global
 ```
 
-## Uninstall
+---
 
-```bash
-npx continuous-improvement uninstall --claude
-npx continuous-improvement uninstall --codex
-npx continuous-improvement uninstall --cursor
-npx continuous-improvement uninstall --openclaw
-```
+## The 7 Laws
 
-## What it installs
+| # | Law | What it prevents |
+|---|-----|-----------------|
+| 1 | **Research before executing** | Reinventing what already exists |
+| 2 | **Plan before coding** | Scope creep and overbuilding |
+| 3 | **One thing at a time** | Stacking untested changes |
+| 4 | **Verify before reporting** | False "done" claims |
+| 5 | **Reflect after sessions** | Repeating the same failures |
+| 6 | **Iterate one change at a time** | Debugging 5 changes at once |
+| 7 | **Learn from every session** | Knowledge that dies with the context window |
 
-Depending on your target, it does one of these:
+---
 
-- **Claude Code** → appends the rules to `CLAUDE.md`
-- **Codex / OpenClaw AGENTS flow** → appends the rules to `AGENTS.md`
-- **Cursor** → appends the rules to `.cursorrules`
-- **OpenClaw** → installs the local skill at `~/.openclaw/skills/continuous-improvement/`
-- **ChatGPT** → prints the exact block to paste into Custom Instructions
+## What happens after install
 
-## Use
-
-One-shot usage is simple:
+Give your agent any task:
 
 > Use continuous-improvement on this task: add a caching layer to my single-server API.
 
-A good response should come back with:
+A disciplined response includes:
 
 1. **Research** — what exists, constraints, risks, simplest path
 2. **Plan** — what will be built
-3. **Anti-scope** — what will not be built
-4. **Verification** — exact checks
+3. **Anti-scope** — what will NOT be built
+4. **Verification** — exact checks to confirm it works
 5. **Fallback** — what to do if it fails
 6. **Reflection** — what to learn after execution
 
-## The 7 rules
+No more "here's 300 lines of code I didn't test."
 
-1. **Research before executing**
-2. **Plan before coding**
-3. **Do one thing at a time**
-4. **Verify before reporting**
-5. **Reflect after non-trivial work**
-6. **Iterate one change at a time**
-7. **Learn from every session**
+---
 
-## What's new in v1.0: Mulahazah
+## Mulahazah: agents that learn (v1.0)
 
-Mulahazah (Arabic: observation) adds instinct-based learning to the loop.
+Mulahazah (Arabic: observation) makes your agent build **instincts** over time.
 
-Agents don't just follow rules — they build instincts over time.
+```
+Session 1:  Agent skips docs → you correct it → instinct created (confidence: 0.3)
+Session 5:  Same pattern observed → confidence rises to 0.6 → agent starts suggesting
+Session 10: Confidence hits 0.8 → agent auto-applies the behavior
+```
 
-**Key features:**
+**How it works:**
 
 - Hooks observe every tool call (<50ms overhead)
-- Instincts carry confidence scores (0.3–0.9 range)
-- Graduated behavior: silent at low confidence, suggest at mid, auto-apply at high
-- Project scoping: instincts can be global or project-specific
+- A background Haiku agent detects patterns and creates instincts
+- Instincts carry confidence scores (0.3–0.9)
+- Graduated behavior: **silent** → **suggest** → **auto-apply**
+- Project-scoped: instincts can be global or per-project
 - Confidence decay: unused instincts weaken, wrong ones get corrected
 
-**One command:**
+**Check what your agent has learned:**
 
 ```bash
 /continuous-improvement
 ```
 
-**Background observer (optional):**
-
-A lightweight Haiku agent watches sessions and extracts patterns automatically.
+**Start the background observer (optional):**
 
 ```bash
 ~/.claude/mulahazah/agents/start-observer.sh
 ```
 
-## Why this exists
-
-Agents usually fail in predictable ways:
-
-- they skip docs and existing code
-- they overbuild
-- they claim success before testing
-- they pile on changes before verifying the first one
-- they repeat mistakes because nothing gets logged
+---
 
-continuous-improvement fixes that with a tiny, reusable operating loop:
-
-```text
-Research → Plan → Execute (one thing) → Verify → Reflect → Learn → Iterate
-```
-
-## Skill first
-
-This repo is designed as a **skill first** product:
-
-- easy to install
-- easy to invoke in one shot
-- useful before any complex task
-
-Prompt variants in `prompts/` are still here, but they are **supporting material**, not the main product.
-
-The OpenClaw skill lives in:
+## Uninstall
 
-```text
-skills/continuous-improvement/SKILL.md
+```bash
+npx continuous-improvement uninstall --claude
+npx continuous-improvement uninstall --codex
+npx continuous-improvement uninstall --cursor
+npx continuous-improvement uninstall --openclaw
 ```
 
-## Testing
-
-Use these files to test whether the install and behavior are actually good:
+---
 
-- `tests/install-checklist.md`
-- `tests/official-evals.md`
-- `tests/prompt-eval.md`
-- `examples/real-test-cases.md`
-- `.github/ISSUE_TEMPLATE/test-report.yml`
+## Manual install
 
-Best practice: have multiple people test across Claude Code, Codex, Cursor, OpenClaw, and ChatGPT, then submit structured reports instead of random comments.
+If you prefer not to use the installer, copy the relevant file into your project:
 
-## Fallback: manual install
+- `prompts/coding-agent.md` — full coding agent prompt
+- `prompts/core.md` — core system prompt
+- `prompts/minimal.md` — 100-word version
+- `skills/continuous-improvement/SKILL.md` — OpenClaw skill
 
-If you do not want the installer, copy the right block manually:
+See `docs/integration-guide.md` for detailed setup.
 
-- `prompts/coding-agent.md`
-- `prompts/core.md`
-- `prompts/minimal.md`
-- `skills/continuous-improvement/SKILL.md`
-
-More setup detail lives in `docs/integration-guide.md`.
+---
 
 ## Quick reality check
 
-If the agent still jumps straight into writing 300 lines without:
+If your agent still jumps straight into writing code without:
 
 - checking what already exists
 - defining anti-scope
 - stating how it will verify success
-- and it doesn't learn from corrections
+- learning from corrections
+
+Then continuous-improvement is not installed properly.
 
-then continuous-improvement is not installed properly.
+---
 
 ## Contributing
 
-Keep it sharp.
-Keep it practical.
-Kill anything that adds friction.
+Keep it sharp. Keep it practical. Kill anything that adds friction.
 
-MIT.
+MIT License.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "continuous-improvement",
-  "version": "1.0.0",
+  "version": "1.1.0",
   "description": "Install structured self-improvement loops with instinct-based learning into Claude Code — research, plan, execute, verify, reflect, learn, iterate.",
   "license": "MIT",
   "bin": {