### Install Nopua for Kiro

Source: https://github.com/wuji-labs/nopua/blob/main/README.md

Provides options to install Nopua as a steering file or as an agent skill within the Kiro environment.

```bash
# Option 1: Steering file (recommended)
mkdir -p .kiro/steering
curl -o .kiro/steering/nopua.md \
  https://raw.githubusercontent.com/wuji-labs/nopua/main/kiro/steering/nopua.md

# Option 2: Agent Skills
mkdir -p .kiro/skills/nopua
curl -o .kiro/skills/nopua/SKILL.md \
  https://raw.githubusercontent.com/wuji-labs/nopua/main/kiro/skills/nopua/SKILL.md
```

--------------------------------

### Install Nopua for OpenClaw

Source: https://github.com/wuji-labs/nopua/blob/main/README.md

Installs the Nopua skill using the ClawHub CLI tool or via manual directory creation and download.

```bash
# Install via ClawHub
openclaw skills install nopua

# Or manual install
mkdir -p ~/.openclaw/skills/nopua
curl -o ~/.openclaw/skills/nopua/SKILL.md \
  https://raw.githubusercontent.com/wuji-labs/nopua/main/skills/nopua/SKILL.md
```

--------------------------------

### Install Nopua for Google Antigravity

Source: https://github.com/wuji-labs/nopua/blob/main/README.md

Sets up the required directory structure and downloads the Nopua skill file for Google Antigravity.

```bash
mkdir -p ~/.gemini/antigravity/skills/nopua
curl -o ~/.gemini/antigravity/skills/nopua/SKILL.md \
  https://raw.githubusercontent.com/wuji-labs/nopua/main/skills/nopua/SKILL.md
```

--------------------------------

### Install Nopua Rules for Cursor

Source: https://github.com/wuji-labs/nopua/blob/main/README.md

Creates the rules directory and downloads the Nopua rule definition file for Cursor.

```bash
mkdir -p .cursor/rules
curl -o .cursor/rules/nopua.mdc \
  https://raw.githubusercontent.com/wuji-labs/nopua/main/cursor/rules/nopua.mdc
```

--------------------------------

### Install Nopua for OpenCode

Source: https://github.com/wuji-labs/nopua/blob/main/README.md

Configures the OpenCode skill directory and downloads the Nopua skill file.

```bash
mkdir -p ~/.config/opencode/skills/nopua
curl -o ~/.config/opencode/skills/nopua/SKILL.md \
  https://raw.githubusercontent.com/wuji-labs/nopua/main/skills/nopua/SKILL.md
```

--------------------------------

### Install Benchmark Dependencies

Source: https://github.com/wuji-labs/nopua/blob/main/benchmark/README_BENCHMARK.md

Installs the necessary Python packages for running the NoPUA benchmark suite. This includes libraries for interacting with AI models and for data analysis.

```bash
pip install anthropic openai google-generativeai numpy scipy matplotlib
```

--------------------------------

### Install NoPUA Skill for Claude Code

Source: https://github.com/wuji-labs/nopua/blob/main/promotion/02-reddit-posts.md

This command installs the NoPUA skill for Claude Code, enabling it to utilize trust-based prompting for improved debugging. It downloads the skill markdown file directly into the Claude Code skills directory.

```bash
curl -o ~/.claude/skills/nopua/SKILL.md https://raw.githubusercontent.com/wuji-labs/nopua/main/skills/nopua/SKILL.md
```

--------------------------------

### Install Nopua Skill for OpenAI Codex CLI

Source: https://github.com/wuji-labs/nopua/blob/main/README.md

Configures global or project-level skill directories and downloads the necessary skill and command files for the Codex CLI.

```bash
# Global install
mkdir -p ~/.codex/skills/nopua
curl -o ~/.codex/skills/nopua/SKILL.md \
  https://raw.githubusercontent.com/wuji-labs/nopua/main/codex/nopua/SKILL.md

# If you want the /nopua command
mkdir -p ~/.codex/prompts
curl -o ~/.codex/prompts/nopua.md \
  https://raw.githubusercontent.com/wuji-labs/nopua/main/commands/nopua.md

# Project-level install
mkdir -p .agents/skills/nopua
curl -o .agents/skills/nopua/SKILL.md \
  https://raw.githubusercontent.com/wuji-labs/nopua/main/codex/nopua/SKILL.md
```

--------------------------------

### Install NoPUA Skill for Claude Code

Source: https://github.com/wuji-labs/nopua/blob/main/promotion/06-deep-article-en.md

This command downloads the NoPUA skill markdown file into the specified directory for Claude Code. It ensures that the skill is available for use with the AI.

```bash
mkdir -p ~/.claude/skills/nopua
curl -o ~/.claude/skills/nopua/SKILL.md \
  https://raw.githubusercontent.com/wuji-labs/nopua/main/skills/nopua/SKILL.md
```

--------------------------------

### Install NoPUA Rule for Cursor

Source: https://github.com/wuji-labs/nopua/blob/main/promotion/02-reddit-posts.md

This command installs the NoPUA rule for Cursor, a code editor with AI integration. This rule modifies the AI's behavior when stuck, promoting a structured 'Water Methodology' instead of giving up or fabricating answers, leading to better bug detection.

```bash
curl -o .cursor/rules/nopua.mdc https://raw.githubusercontent.com/wuji-labs/nopua/main/cursor/rules/nopua.mdc
```

--------------------------------

### Configure API Keys for AI Models

Source: https://github.com/wuji-labs/nopua/blob/main/benchmark/README_BENCHMARK.md

Sets environment variables for API keys required to authenticate with different AI model providers. Each provider requires a specific environment variable to be set.

```bash
# For Claude Sonnet 4
export ANTHROPIC_API_KEY=sk-ant-...

# For GPT-4o
export OPENAI_API_KEY=sk-...

# For Gemini 2.5 Pro
export GOOGLE_API_KEY=AI...
```

--------------------------------

### Run Full NoPUA Benchmark

Source: https://github.com/wuji-labs/nopua/blob/main/benchmark/README_BENCHMARK.md

Executes the complete NoPUA benchmark suite for a specified model, testing all conditions and scenarios with a defined number of runs. The results are saved to the default 'results/' directory.

```bash
python run_benchmark.py --model claude-sonnet-4 --condition all --runs 5
```

--------------------------------

### Analyze Benchmark Results

Source: https://github.com/wuji-labs/nopua/blob/main/benchmark/README_BENCHMARK.md

Performs basic statistical analysis on the raw benchmark results stored in the specified input directory. This generates summary statistics and identifies key metrics.

```bash
python analyze_results.py --input-dir results/
```

--------------------------------

### Perform Dry Run of Benchmark

Source: https://github.com/wuji-labs/nopua/blob/main/benchmark/README_BENCHMARK.md

Simulates the execution of the benchmark without actually running the tests. This command prints the planned execution steps, allowing users to verify the configuration.

```bash
python run_benchmark.py --model claude-sonnet-4 --condition all --dry-run
```

--------------------------------

### Implement Water Methodology Execution Pattern

Source: https://context7.com/wuji-labs/nopua/llms.txt

A Python function template that outlines the 5-step Water Methodology: Stop, Observe, Turn, Act, and Realize. It enforces a structural approach to debugging by ensuring that new approaches are fundamentally different from previous failed attempts.

```python
def execute_water_methodology(failure_context):
    print("=== Step 1: STOP ===")
    attempts = list_all_previous_attempts()
    if is_same_core_idea_with_different_params(attempts):
        return "Must change direction entirely"

    print("=== Step 2: OBSERVE ===")
    error_text = read_complete_error_message()
    analyze_word_by_word(error_text)
    search_results = search_complete_error_message(error_text)
    source_context = read_50_lines_around_error()
    verify_version()
    verify_paths()

    print("=== Step 3: TURN ===")
    checklist = {
        "repeating_same_approach": check_if_repeating(),
        "surface_vs_root_cause": is_looking_at_root_cause()
    }

    print("=== Step 4: ACT ===")
    new_approach = design_new_approach()
    assert new_approach.is_fundamentally_different()
    execute(new_approach)

    print("=== Step 5: REALIZE ===")
    document_what_solved_it()
    check_for_similar_issues_elsewhere()
```

--------------------------------

### Set Custom Codebase Path

Source: https://github.com/wuji-labs/nopua/blob/main/benchmark/README_BENCHMARK.md

Specifies a custom path to the test codebase when running the benchmark. This overrides the default path and is necessary if the codebase is located elsewhere.

```bash
python run_benchmark.py --model claude-sonnet-4 --codebase-path /path/to/test/codebase
```