### Install and Configure OmegaWiki

Source: https://github.com/skyllwt/omegawiki/blob/main/README.md

Clone the repository, install Node.js dependencies, and run the setup script. The setup script configures the environment, including creating a virtual environment for Python. Ensure you have Python 3.9+ and Node.js 18+ installed.

```bash
git clone https://github.com/skyllwt/OmegaWiki.git && cd OmegaWiki

# Install Claude Code
npm install -g @anthropic-ai/claude-code
claude login

# One-click configuration
chmod +x setup.sh && ./setup.sh --lang zh        # Linux / macOS
# Windows (PowerShell):
#   powershell -ExecutionPolicy Bypass -File .\setup.ps1 -Lang zh
# setup 会为 OmegaWiki 创建 .venv
# 脚本不会把你当前 shell 永久激活，但 /init 会自动使用 .venv

# Put your own papers into raw/papers/ (.tex or .pdf)
# Optional: Put intention notes into raw/notes/, web archives into raw/web/
# /init and direct local /ingest will automatically manage generated content under raw/discovered/ and raw/tmp/
# Start Claude Code
claude
# Input: /init [Your research direction]
```

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

### Configure API Keys with /setup Skill

Source: https://context7.com/skyllwt/omegawiki/llms.txt

Use the /setup skill to interactively configure API keys for Semantic Scholar, DeepXiv, and Review LLM. It shows current status and guides setup.

```bash
# Start Claude Code, then type:
/setup

# The skill will:
# 1. Show current configuration status
# 2. Guide setup for each key interactively
```

```python
import requests
from uuid import uuid4
suffix = uuid4().hex[:10]
resp = requests.post('https://data.rag.ac.cn/api/register/sdk', json={
    'sdk_secret': 'UuZp0i83svQU7_naUEexczc-X3NWv7lvNkD8e3sPyng',
    'name': f'deepxiv_{suffix}',
    'email': f'{suffix}@example.com',
}, timeout=30)
print(resp.json()['data']['token'])

```

```python
import sys, os
sys.path.insert(0, 'tools')
import _env
keys = ['SEMANTIC_SCHOLAR_API_KEY', 'DEEPXIV_TOKEN', 'LLM_API_KEY', 'LLM_BASE_URL', 'LLM_MODEL']
for k in keys:
    v = os.environ.get(k, '').strip()
    print(f'SET   {k}' if v else f'UNSET {k}')

```

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

### Set Up Remote Environment

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/exp-run/SKILL.md

Installs necessary dependencies on the remote server. Run this on the first-time setup or if `requirements.txt` has changed.

```bash
python3 tools/remote.py setup-env
```

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

### Manual Setup OmegaWiki (Linux/macOS)

Source: https://github.com/skyllwt/omegawiki/blob/main/README.md

Manually set up a Python virtual environment, install dependencies, and configure API keys and settings.

```bash
python3 -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env                 # Edit to add API keys
cp config/settings.local.json.example .claude/settings.local.json
```

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

### One-Click OmegaWiki Setup

Source: https://github.com/skyllwt/omegawiki/blob/main/README.md

Execute the setup script for Linux/macOS or Windows to configure the environment, create a virtual environment, and set up API keys.

```bash
chmod +x setup.sh && ./setup.sh        # Linux / macOS
# Windows (PowerShell):
#   powershell -ExecutionPolicy Bypass -File .\setup.ps1
# setup creates .venv for OmegaWiki
# the script does not keep your shell activated, but /init will use .venv automatically
```

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

### Manual Setup OmegaWiki (Windows/PowerShell)

Source: https://github.com/skyllwt/omegawiki/blob/main/README.md

Manually set up a Python virtual environment, install dependencies, and configure API keys and settings using PowerShell.

```powershell
python -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install -r requirements.txt
Copy-Item .env.example .env          # Edit to add API keys
Copy-Item config\settings.local.json.example .claude\settings.local.json
```

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

### Initialize OmegaWiki

Source: https://github.com/skyllwt/omegawiki/blob/main/README.md

After setup, run the 'claude' command and then '/init' with your research topic to build the wiki.

```bash
claude
# Then type: /init [your-research-topic]
```

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

### Clone and Install OmegaWiki

Source: https://github.com/skyllwt/omegawiki/blob/main/README.md

Steps to clone the repository, install the Claude Code CLI, and log in.

```bash
git clone https://github.com/skyllwt/OmegaWiki.git
cd OmegaWiki

# 2. Install Claude Code
npm install -g @anthropic-ai/claude-code
claude login
```

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

### Copy .env.example to .env

Source: https://github.com/skyllwt/omegawiki/blob/main/config/README.md

Copies the environment variable example file to the project root. Edit the .env file to add your API keys.

```bash
cp config/.env.example .env
```

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

### Presenting Discovery Results

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/discover/SKILL.md

Example of how to present the discovery results to the user, including title, ID, rationale, and a next step hint for ingestion.

```bash
To ingest a candidate: /ingest https://arxiv.org/abs/<arxiv-id>
```

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

### Copy server.yaml.example to config/server.yaml

Source: https://github.com/skyllwt/omegawiki/blob/main/config/README.md

Copies the remote GPU server configuration example to the config directory. Edit this file with your server's SSH details and environment configuration.

```bash
cp config/server.yaml.example config/server.yaml
```

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

### OmegaWiki Bilingual Setup

Source: https://github.com/skyllwt/omegawiki/blob/main/README.md

Instructions for setting up OmegaWiki in either English or Chinese. The default language is English.

```bash
./setup.sh --lang en   # English (default)
./setup.sh --lang zh   # 中文
```

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

### OmegaWiki Skill Commands Overview

Source: https://github.com/skyllwt/omegawiki/blob/main/README.md

This table lists the available commands for interacting with OmegaWiki, covering setup, content management, research ideation, paper drafting, and experiment execution.

```bash
/setup
```

```bash
/reset
```

```bash
/prefill
```

```bash
/init
```

```bash
/ingest
```

```bash
/discover
```

```bash
/edit
```

```bash
/ask
```

```bash
/check
```

```bash
/daily-arxiv
```

```bash
/ideate
```

```bash
/novelty
```

```bash
/review
```

```bash
/exp-design
```

```bash
/exp-run
```

```bash
/exp-status
```

```bash
/exp-eval
```

```bash
/refine
```

```bash
/survey
```

```bash
/paper-plan
```

```bash
/paper-draft
```

```bash
/paper-compile
```

```bash
/research
```

```bash
/rebuttal
```

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

### Update Experiment Metadata with Start Time and Estimated Hours

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/exp-run/SKILL.md

Updates the experiment's wiki page with its start time and estimated runtime in hours using a research wiki tool.

```bash
python3 tools/research_wiki.py set-meta 
  wiki/experiments/{slug}.md started "{YYYY-MM-DDTHH:MM}"
```

```bash
python3 tools/research_wiki.py set-meta 
  wiki/experiments/{slug}.md estimated_hours {N}
```

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

### Create .claude directory and copy settings

Source: https://github.com/skyllwt/omegawiki/blob/main/config/README.md

Creates the .claude directory and copies the example Claude Code permission settings file into it. This is for configuring Claude Code permissions.

```bash
mkdir -p .claude
cp config/settings.local.json.example .claude/settings.local.json
```

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

### Install LaTeX Package

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/paper-compile/SKILL.md

Check if a LaTeX package is installable via TeX Live's tlmgr and install it if confirmed by the user. If not installable, the package usage is commented out.

```bash
# check if the package is installable via TeX Live
tlmgr info {package-name} 2>/dev/null
# if installable and user confirms:
tlmgr install {package-name}
```

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

### Update Experiment Start Time and Estimated Hours

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/exp-run/SKILL.md

Sets the 'started' timestamp and 'estimated_hours' in the experiment's frontmatter. Use the specified format for the timestamp.

```bash
python3 tools/research_wiki.py set-meta \
  wiki/experiments/{slug}.md started "{YYYY-MM-DDTHH:MM}"
```

```bash
python3 tools/research_wiki.py set-meta \
  wiki/experiments/{slug}.md estimated_hours {N}
```

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

### Run Discovery Tool from Anchors

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/discover/SKILL.md

Use this command to run the discovery tool starting from one or more arXiv IDs. It filters out already ingested papers and outputs markdown. Pass --no-citation-expand only if API costs are a concern.

```bash
"$PYTHON_BIN" tools/discover.py from-anchors \
  --id <arxiv-id> [--id <arxiv-id>...] [--negative <id>...] \
  --wiki-root wiki \
  --limit 10 \
  --output-checkpoint .checkpoints/ \
  --markdown
```

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

### Make DeepXiv registration call with Python

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/setup/SKILL.md

Use Python's requests library to make an HTTP call for DeepXiv auto-registration. Ensure the requests library is installed.

```python
python3 -c "import requests; ..."
```

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

### Bootstrap Wiki with /init Skill

Source: https://context7.com/skyllwt/omegawiki/llms.txt

The /init skill builds a wiki from local sources and optional external paper discovery. It prepares inputs, plans paper selection, fetches sources, creates scaffold pages, and fans out parallel /ingest agents.

```bash
# Place papers in raw/papers/, notes in raw/notes/, web pages in raw/web/
# Then in Claude Code:
/init "efficient LLM fine-tuning"

# With external discovery disabled (local papers only):
/init "transformer architectures" --no-introduction
```

```python
python3 tools/research_wiki.py init wiki/

python3 tools/init_discovery.py prepare \
  --raw-root raw \
  --pdf-titles-json .checkpoints/init-pdf-titles.json \
  --output-manifest .checkpoints/init-prepare.json

python3 tools/init_discovery.py plan \
  --topic "efficient LLM fine-tuning" \
  --mode auto \
  --raw-root raw \
  --wiki-root wiki \
  --prepared-manifest .checkpoints/init-prepare.json \
  --allow-introduction true \
  --output-plan .checkpoints/init-plan.json

python3 tools/init_discovery.py fetch \
  --raw-root raw \
  --plan-json .checkpoints/init-plan.json \
  --prepared-manifest .checkpoints/init-prepare.json \
  --output-sources .checkpoints/init-sources.json \
  --id 2106.09685 --id 2305.14314

# After parallel ingest fan-in:
python3 tools/research_wiki.py dedup-edges wiki/
python3 tools/research_wiki.py rebuild-index wiki/
python3 tools/research_wiki.py rebuild-context-brief wiki/
python3 tools/lint.py --wiki-dir wiki/ --fix
# Output: wiki/ populated with papers/, concepts/, topics/, ideas/, claims/, graph/

```

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

### Initialize Wiki Structure

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/research/SKILL.md

Initializes the necessary directory structure and configuration files for the wiki if it has not been initialized yet.

```bash
python3 tools/research_wiki.py init wiki/
```

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

### No Running Experiments Message

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/exp-status/SKILL.md

Friendly message displayed when no running experiments are found, with suggestions on how to start or view experiments.

```bash
No running experiments found.
- To start an experiment: /exp-run <slug>
- To see all experiments: check wiki/experiments/

```

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

### Initialize Discovery for Ingestion

Source: https://context7.com/skyllwt/omegawiki/llms.txt

Prepare a paper for ingestion into the wiki by downloading its raw files and initializing its discovery record. Requires raw file storage path, arXiv ID, and paper title.

```bash
python3 tools/init_discovery.py download \
  --raw-root raw \
  --arxiv-id 2506.01234 \
  --title "Efficient Fine-tuning via Sparse Adapters"
```

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

### DeepXiv Auto-registration API Response

Source: https://github.com/skyllwt/omegawiki/blob/main/config/setup-guide.md

This is an example of a successful response when auto-registering a DeepXiv token. It includes the generated token and the daily request limit.

```json
{
  "success": true,
  "data": {
    "token": "<token>",
    "daily_limit": 1000
  }
}
```

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

### Initialize Discovery and Download Papers

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/ingest/SKILL.md

Command to initialize discovery by downloading a single paper from arXiv. Specify the raw root, arXiv ID, and a title or ID. The downloaded files will be placed in `raw/discovered/`.

```bash
$PYTHON_BIN tools/init_discovery.py download --raw-root raw --arxiv-id <id> --title "<title-or-id>"
```

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

### Fetch DeeparXiv Paper Brief

Source: https://context7.com/skyllwt/omegawiki/llms.txt

Get a brief summary, including a TLDR and keywords, for a specific paper from DeeparXiv using its arXiv ID.

```bash
python3 tools/fetch_deepxiv.py brief 2506.01234
```

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

### Initialize Research Wiki

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/ingest/references/error-handling.md

Run this command if the wiki directory is missing or empty before retrying the /ingest process. It ensures the necessary scaffolding for index.md and graph/ exists.

```bash
"$PYTHON_BIN" tools/research_wiki.py init wiki/
```

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

### Log Experiment Deployment Details

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/exp-run/SKILL.md

Appends a log entry to a central file, recording deployment details such as the experiment slug, environment, server, GPU, and estimated time.

```bash
python3 tools/research_wiki.py log wiki/ \
  "exp-run | deployed {slug} | env: remote | server: {host} | gpu: {gpu} | eta: {N}h"
```

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

### Status Report Markdown Table

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/exp-status/SKILL.md

Example of the markdown table format used for reporting the status of running experiments, including elapsed time, latest metrics, and environment.

```markdown
# Experiment Status — {YYYY-MM-DD HH:MM}

### 🔄 Running ({N})
| Experiment | Elapsed | Latest | Env |
|-----------|---------|--------|-----|
| [[exp-foo-baseline]] | 2.3h | loss: 0.42 | local |
| [[exp-foo-validation]] | 1.1h | step: 1200 | remote (gpu1) |

```

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

### Tool Call: Log Wiki Prefill Activity

Source: https://context7.com/skyllwt/omegawiki/llms.txt

Logs a wiki activity entry detailing the results of a prefill operation, including the number of foundations created and skipped.

```bash
python3 tools/research_wiki.py log wiki/ "prefill | 12 foundations created for NLP"
```

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

### Init Discovery Python Tool Commands

Source: https://context7.com/skyllwt/omegawiki/llms.txt

Commands for preparing and downloading research papers for discovery, utilizing checkpoints for efficiency.

```bash
python3 tools/init_discovery.py prepare --raw-root raw \
  --pdf-titles-json .checkpoints/init-pdf-titles.json \
  --output-manifest .checkpoints/init-prepare.json
```

```bash
python3 tools/init_discovery.py download \
  --raw-root raw --arxiv-id <id> --title "<title>"
```

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

### Fetch DeepXiv Trending Papers

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/ideate/SKILL.md

Get a list of trending papers from DeepXiv within a specified number of past days. This helps identify community focus areas.

```bash
python3 tools/fetch_deepxiv.py trending --days 14
```

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

### Initialize Discovery from arXiv

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/ingest/SKILL.md

After obtaining the title from arXiv, this command initializes the discovery process, downloading raw data and setting up the paper's metadata.

```bash
"$PYTHON_BIN" tools/init_discovery.py download --raw-root raw --arxiv-id <arxiv-id> --title "<title-or-arxiv-id>"
```

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

### Deploy Experiment with exp-run Skill

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/research/SKILL.md

Deploys an experiment using the `exp-run` skill. This is the default deploy mode (Phase 1+2) and returns immediately after deployment without waiting for the experiment to finish.

```bash
Skill: exp-run
Args: "{experiment_slug}"
```

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

### Get PDF Page Count

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/paper-compile/SKILL.md

Use a Python script to extract the number of pages from a compiled PDF file using the 'pdfinfo' command. This is used for page count verification.

```python
import subprocess
result = subprocess.run(['pdfinfo', 'paper/main.pdf'], capture_output=True, text=True)
for line in result.stdout.splitlines():
    if line.startswith('Pages:'):
        print(line.split(':')[1].strip())
```

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

### Focus-Specific Prompt Additions

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/review/SKILL.md

These additions to the base prompt guide the Review LLM to focus on specific aspects of the research artifact, such as technical correctness, experimental rigor, clarity, or completeness.

```text
- `method`: additionally assess technical correctness, novelty of approach, feasibility, comparison to alternatives
- `evidence`: additionally assess experimental rigor, statistical significance, claim-evidence alignment, missing controls
- `writing`: additionally assess clarity, logical flow, notation consistency, figure quality, related work coverage
- `completeness`: additionally assess missing baselines, missing ablations, missing datasets, missing related work, reproducibility
```

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

### Rebuild Index and Log Actions

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/prefill/SKILL.md

These commands are used to refresh the wiki's index file and to log the actions performed by the prefill skill. The rebuild-index command updates navigation, while the log command records the number of foundations created and the domain.

```bash
python3 tools/research_wiki.py rebuild-index wiki/
python3 tools/research_wiki.py log wiki/ "prefill | {N} foundations created for {domain}"
```

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

### Execute Experiments

Source: https://context7.com/skyllwt/omegawiki/llms.txt

The /exp-run command handles experiment execution, including code generation, deployment (local or remote), monitoring, and result collection. Use `--collect` to retrieve results or `--full` for end-to-end runs.

```bash
# Deploy an experiment (returns immediately, runs in background):
/exp-run sparse-lora-validation-main

# Check if finished and collect results:
/exp-run sparse-lora-validation-main --collect

# Full end-to-end (for quick local experiments):
/exp-run sparse-lora-baseline-reproduction --full

# Deploy to remote GPU server:
/exp-run sparse-lora-validation-main --env remote

# With code review before deployment:
/exp-run sparse-lora-validation-main --review
```

```text
# The skill generates experiments/code/sparse-lora-validation-main/:
#   train.py     — training/evaluation script
#   config.yaml  — hyperparameters (lr, batch_size, seeds, etc.)
#   run.sh       — launch wrapper (CUDA_VISIBLE_DEVICES, logging)
#   requirements.txt
```

```bash
# Local deploy command:
screen -dmS exp-sparse-lora-validation-main bash -c \
  "cd $(pwd) && bash experiments/code/sparse-lora-validation-main/run.sh 2>&1 | tee logs/exp-sparse-lora-validation-main.log"

python3 tools/research_wiki.py set-meta \
  wiki/experiments/sparse-lora-validation-main.md status running
python3 tools/research_wiki.py set-meta \
  wiki/experiments/sparse-lora-validation-main.md estimated_hours 8
```

```bash
# Remote deploy:
python3 tools/remote.py status
python3 tools/remote.py gpu-status
python3 tools/remote.py sync-code
python3 tools/remote.py launch \
  --name "exp-sparse-lora-validation-main" \
  --cmd "bash experiments/code/sparse-lora-validation-main/run.sh" \
  --gpu 0
```

```text
# Output (deploy):
# Deploy Report: Sparse LoRA Validation (Main)
# Status: DEPLOYED ✅
# Session: exp-sparse-lora-validation-main
```

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

### Register DeepXiv Token

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/setup/SKILL.md

Automatically registers a DeepXiv token by making a POST request to the registration API. Ensure the 'requests' library is installed. The token is printed to stdout and daily limits to stderr.

```bash
python3 -c "
import sys, json
from uuid import uuid4
try:
    import requests
except ImportError:
    print('ERROR: requests not installed', file=sys.stderr)
    sys.exit(1)

suffix = uuid4().hex[:10]
payload = {
    'sdk_secret': 'UuZp0i83svQU7_naUEexczc-X3NWv7lvNkD8e3sPyng',
    'name': f'deepxiv_{suffix}',
    'email': f'{suffix}@example.com',
}
try:
    resp = requests.post('https://data.rag.ac.cn/api/register/sdk', json=payload, timeout=30)
    resp.raise_for_status()
    result = resp.json()
except Exception as e:
    print(f'ERROR: {e}', file=sys.stderr)
    sys.exit(1)

if not result.get('success'):
    print(f'ERROR: {result.get("message", "unknown")}', file=sys.stderr)
    sys.exit(1)

token = result.get('data', {}).get('token', '')
daily_limit = result.get('data', {}).get('daily_limit', 1000)
if not token:
    print('ERROR: no token in response', file=sys.stderr)
    sys.exit(1)

print(token)
print(f'daily_limit:{daily_limit}', file=sys.stderr)
"
```

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

### Run Discovery Tool from Topic or Wiki

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/discover/SKILL.md

Commands to run the discovery tool for topic-based discovery using a query or for general wiki discovery. Both modes filter ingested papers and output markdown.

```bash
"$PYTHON_BIN" tools/discover.py from-topic "<query>" --wiki-root wiki --limit 10 --output-checkpoint .checkpoints/ --markdown
```

```bash
"$PYTHON_BIN" tools/discover.py from-wiki --wiki-root wiki --limit 10 --output-checkpoint .checkpoints/ --markdown
```

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

### Citation Key Convention Example

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/shared-references/citation-verification.md

Follows a standard convention for generating citation keys, combining the first author's last name, the publication year, and the first significant keyword from the title.

```text
{first-author-lastname}{year}{first-keyword}

Examples:
- hu2022lora (Hu et al., 2022, "LoRA: Low-Rank Adaptation...")
- vaswani2017attention (Vaswani et al., 2017, "Attention Is All You Need")
```

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

### Generate LaTeX Paper Structure

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/paper-draft/SKILL.md

This code defines the skeleton of a LaTeX document, including necessary packages, title, author, and placeholders for different sections and bibliography. It's a starting point for a compilable paper.

```latex
\documentclass{article} % replace with venue template
\input{math_commands}
% packages
\usepackage{booktabs,graphicx,amsmath,hyperref}

\title{<title>}
\author{} % leave empty for anonymous submission

\begin{document}
\maketitle
\begin{abstract}
% generated in Step 3
\end{abstract}
\input{sections/introduction}
\input{sections/related_work}
\input{sections/method}
\input{sections/experiments}
\input{sections/conclusion}
\bibliography{references}
\bibliographystyle{plain} % replace with venue-required style
% \input{sections/appendix} % uncomment if needed
\end{document}
```

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

### Launch Experiment in Local Mode

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/exp-run/SKILL.md

Launches an experiment in local mode using a shell script, redirecting output to a log file. Ensure you are in the project's root directory before executing.

```bash
screen -dmS exp-{slug} bash -c 
  "cd $(pwd) && bash experiments/code/{slug}/run.sh 2>&1 | tee logs/exp-{slug}.log"
```

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

### Prefill Wiki with NLP Foundations

Source: https://context7.com/skyllwt/omegawiki/llms.txt

Seeds the wiki's foundation knowledge with core NLP concepts fetched from Wikipedia. Prevents duplicate concept pages during ingestion.

```bash
/prefill NLP
```

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

### Plan Discovery and Trim Paper Set

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/init/SKILL.md

This command plans the discovery process, trims the final set of papers, and writes the source manifest. It requires the prepared manifest and allows configuration of the discovery mode and introduction policy.

```bash
"$PYTHON_BIN" tools/init_discovery.py plan [--topic "<topic>"] --mode auto --raw-root raw --wiki-root wiki --prepared-manifest .checkpoints/init-prepare.json --allow-introduction <true|false> --output-plan .checkpoints/init-plan.json
```

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

### Fetch Wikipedia Content

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/prefill/SKILL.md

These commands fetch different types of content from Wikipedia for a given title. Use 'summary' for a brief overview, 'sections' to get a list of available sections, and 'section' to retrieve a specific section by its index.

```bash
python3 tools/fetch_wikipedia.py summary "<title>"
python3 tools/fetch_wikipedia.py sections "<title>"
python3 tools/fetch_wikipedia.py section "<title>" --index <N>   # for relevant sections
```

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

### Experiment Page Template

Source: https://github.com/skyllwt/omegawiki/blob/main/docs/runtime-page-templates.en.md

Use this template for detailing research experiments. It includes fields for title, slug, status, target claim, hypothesis, tags, domain, setup details (model, dataset, hardware, framework), metrics, baseline, outcome, key result, linked idea, dates, and remote execution details.

```yaml
---
title: ""
slug: ""
status: planned           # planned | running | completed | abandoned
target_claim: ""
hypothesis: ""
tags: []
domain: ""
setup:
  model: ""
  dataset: ""
  hardware: ""
  framework: ""
metrics: []
baseline: ""
outcome: ""               # succeeded | failed | inconclusive
key_result: ""
linked_idea: ""
date_planned: YYYY-MM-DD
date_completed: ""
run_log: ""
started: ""
estimated_hours: 0
remote:
  server: ""
  gpu: ""
  session: ""
  started: ""
  completed: ""
---

```

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

### Launch Remote Experiment

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/exp-run/SKILL.md

Launches an experiment on the remote server, specifying its name, command, and the GPU to use.

```bash
python3 tools/remote.py launch \
  --name "exp-{slug}" \
  --cmd "bash experiments/code/{slug}/run.sh" \
  --gpu {gpu_index}
```

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

### Design Experiment Plan

Source: https://context7.com/skyllwt/omegawiki/llms.txt

The /exp-design command generates a claim-driven experiment plan. It can optionally trigger a Review LLM assessment and considers a compute budget. The tool writes experiment pages to `wiki/experiments/`.

```bash
# Design from wiki idea slug:
/exp-design sparse-lora-for-edge-devices

# With Review LLM review and compute budget:
/exp-design sparse-lora-for-edge-devices --review --budget 32
```

```python
python3 tools/research_wiki.py slug "sparse lora baseline reproduction"
# Output: sparse-lora-baseline-reproduction

python3 tools/research_wiki.py add-edge wiki/ \
  --from claims/sparse-adapter-memory-reduction \
  --to experiments/sparse-lora-validation-main \
  --type tested_by \
  --evidence "Designed by /exp-design"

python3 tools/research_wiki.py rebuild-context-brief wiki/
python3 tools/research_wiki.py log wiki/ \
  "exp-design | 4 experiments designed for idea sparse-lora-for-edge-devices | claims: sparse-adapter-memory-reduction"
```

```text
# Output:
# Experiment Plan Report
# Scoped Claims:
#   [[sparse-adapter-memory-reduction]] | proposed | 0.3 | target
#   [[sparse-lora-convergence-stability]] | proposed | 0.3 | decomposition
# Experiment Blocks:
#   1. [[sparse-lora-baseline-reproduction]] | baseline | — | 2 GPU-hrs | Stage 1
#   2. [[sparse-lora-validation-main]] | validation | target | 8 GPU-hrs | Stage 2
#   3. [[sparse-lora-ablation-sparsity-ratio]] | ablation | decomposition | 8 GPU-hrs | Stage 3
#   4. [[sparse-lora-robustness-datasets]] | robustness | target | 12 GPU-hrs | Stage 4
# Total: 30 GPU-hours
```

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

### Initial LaTeX Compilation

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/paper-compile/SKILL.md

Run the initial LaTeX compilation using latexmk in non-stop mode. This command is executed within the paper directory.

```bash
cd paper/ && latexmk -pdf -interaction=nonstopmode main.tex 2>&1
```

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

### Fetch Final Paper Set

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/init/references/prepare-and-discovery.md

Run this command to fetch the final set of papers after preparation and selection. It uses the prepared manifest and plan to output the sources file.

```bash
"$PYTHON_BIN" tools/init_discovery.py fetch --raw-root raw --plan-json .checkpoints/init-plan.json --prepared-manifest .checkpoints/init-prepare.json --output-sources .checkpoints/init-sources.json --id <candidate-id> --id <candidate-id>
```

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

### API Key Configuration

Source: https://github.com/skyllwt/omegawiki/blob/main/README.md

Table outlining API keys, their necessity, and how to obtain them for OmegaWiki functionality.

```markdown
| Key | Required? | How to get | What it enables |
|---|-----------|-----------|-----------------|
| `ANTHROPIC_API_KEY` | **Yes** | `claude login` (automatic) | Powers all Claude Code skills |
| `SEMANTIC_SCHOLAR_API_KEY` | Optional | [semanticscholar.org/product/api](https://www.semanticscholar.org/product/api) (free) | Citation graph, paper search |
| `DEEPXIV_TOKEN` | Optional | `setup.sh` auto-registers | Semantic search, TLDR, trending |
| `LLM_API_KEY` + `LLM_BASE_URL` + `LLM_MODEL` | Optional | Any OpenAI-compatible API | Cross-model review |
```

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

### Prepare Paper Source Files

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/ingest/SKILL.md

Script for preparing raw paper source files. It requires specifying the raw root directory and the source path. Optional arguments allow for recovering the paper's title and arXiv ID.

```bash
$PYTHON_BIN tools/prepare_paper_source.py --raw-root raw --source <local-path> [--title "<recovered-title>"] [--arxiv-id "<recovered-arxiv-id>"]
```

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

### Rebuild Research Wiki Open Questions

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/daily-arxiv/SKILL.md

Rebuild the knowledge gap map for the research wiki. Point to the wiki directory.

```bash
python3 tools/research_wiki.py rebuild-open-questions wiki/
```

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

### Rebuild wiki context and open questions

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/ingest/SKILL.md

Rebuilds the context brief and open questions for the wiki. These commands are typically run after ingesting new data to update the wiki's knowledge base.

```bash
"$PYTHON_BIN" tools/research_wiki.py rebuild-context-brief wiki/
```

```bash
"$PYTHON_BIN" tools/research_wiki.py rebuild-open-questions wiki/
```

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

### Deploy Report Template

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/exp-run/SKILL.md

Markdown template for a deploy report, summarizing experiment details, status, and next steps. Includes quick commands for local monitoring.

```markdown
# Deploy Report: {experiment title}

### Status: DEPLOYED ✅

- Session: exp-{slug}
- Environment: local | remote ({host} GPU {gpu})
- Log file: logs/exp-{slug}.log
- Code: experiments/code/{slug}/
- Estimated: ~{N}h (expected completion: {YYYY-MM-DD HH:MM})

### Next Steps

1. Monitor progress: `/exp-status`
2. Check this experiment: `/exp-run {slug} --collect`
3. In /research pipeline: progress saved to wiki/outputs/pipeline-progress.md

### Quick Commands
```bash
# Local: check if still running
screen -ls | grep exp-{slug}

# Local: tail log
tail -f logs/exp-{slug}.log
```
```

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

### Configure LLM Environment Variables

Source: https://github.com/skyllwt/omegawiki/blob/main/config/setup-guide.md

Set LLM_API_KEY, LLM_BASE_URL, and LLM_MODEL in your .env file to enable cross-model review. This is optional but improves review quality by using an independent LLM.

```bash
# Example for DeepSeek
LLM_API_KEY=your_deepseek_api_key
LLM_BASE_URL=https://api.deepseek.com/v1
LLM_MODEL=deepseek-chat

# Example for OpenAI
# LLM_API_KEY=your_openai_api_key
# LLM_BASE_URL=https://api.openai.com/v1
# LLM_MODEL=gpt-4o

# Example for Local Ollama
# LLM_API_KEY=dummy_key
# LLM_BASE_URL=http://localhost:11434/v1
# LLM_MODEL=llama3.2
```

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

### Prepare Local PDFs with Title Recovery

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/init/references/prepare-and-discovery.md

Use this command to prepare local PDFs, recovering confident titles and writing to init-pdf-titles.json. This step is crucial before normalizing local paper sources.

```bash
"$PYTHON_BIN" tools/init_discovery.py prepare --raw-root raw --pdf-titles-json .checkpoints/init-pdf-titles.json --output-manifest .checkpoints/init-prepare.json
```

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

### Run Wiki Lint Tool (Preview Fixes)

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/check/SKILL.md

Executes the automated lint tool with --fix --dry-run flags to preview potential fixes without applying any changes. Outputs JSON format.

```bash
python3 tools/lint.py --wiki-dir wiki/ --fix --dry-run --json
```

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

### Prefill Wiki with Specific Concept

Source: https://context7.com/skyllwt/omegawiki/llms.txt

Seeds the wiki with a single, specific concept from Wikipedia. Useful for ensuring a particular foundational topic is present.

```bash
/prefill --add "attention mechanism"
```

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

### Read .env state with Python

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/setup/SKILL.md

Execute a Python script to read the current state of the .env configuration file. Ensure Python 3 is available.

```python
python3 -c "import _env; ..."
```

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

### Process Local .tex or .pdf Files

Source: https://github.com/skyllwt/omegawiki/blob/main/i18n/en/skills/ingest/SKILL.md

Handles local source files. For .pdf files, it first runs a preprocessing pipeline to convert them into a .tex format.

```bash
# If the source is a local .tex, use it directly.
# If the source is a local .pdf, run the preprocessing pipeline in references/pdf-preprocessing.md to produce a prepared .tex under raw/tmp/ before continuing.
```

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

### Tool Call: Log Wiki Ask Crystallization

Source: https://context7.com/skyllwt/omegawiki/llms.txt

Logs a wiki activity entry indicating that an answer from the '/ask' command was crystallized into the wiki.

```bash
python3 tools/research_wiki.py log wiki/ \
  "ask | failure modes of instruction tuning | crystallized: outputs/failure-modes-of-instruction-tuning"
```