Tavily codex

Skill azel-ko/tavily-codex-skill/skills/tavily-codex

Tavily Agent Skill for OpenAI Codex and Claude Code: web search, extract, crawl, map, and research APIs without the Tavily CLI

Install

npx -y skills add azel-ko/tavily-codex-skill --skill tavily-codex

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

23 days oldThe repository was created 23 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use Tavily web search, extract, crawl, map, research, and usage APIs from Codex or compatible Agent Skills clients through bundled Python helpers. Use when Codex needs current web research, cited source discovery, URL content extraction, site crawling, URL mapping, Tavily API payload dry-runs, or Tavily authentication guidance without relying on the Tavily CLI.

SKILL.md

3.1 KB, as published. Nobody here has run it

Tavily Codex

Overview

Use this skill to call Tavily's REST API directly from Codex with stable JSON output. The bundled helper uses Python standard library only and does not require the official tvly CLI.

Workflow

Read references/authentication.md when credentials or API key handling are relevant.
Choose the lightest operation that satisfies the request:
- search: find pages or answer a current-information question.
- extract: pull clean content from known URLs.
- map: discover URLs on a site before choosing pages.
- crawl: extract many pages from a site section.
- research: start a multi-source research task with citations.
- research-status: fetch the status/result for a research request ID.
- usage: check account/key usage.
Run a dry-run first unless the user already supplied an exact command and explicitly asked to execute it.
Add --execute only when the target, expected cost/side effect, and credentials are clear.

Helper

Run the helper from this skill folder:

python3 scripts/tavily_api.py search --query "latest OpenAI API changes" --max-results 5
python3 scripts/tavily_api.py search --query "latest OpenAI API changes" --max-results 5 --execute

If the skill is installed in Claude Code, use:

python3 "$CLAUDE_SKILL_DIR/scripts/tavily_api.py" search --query "..." --execute

The helper reads TAVILY_API_KEY from the environment, redacts token-shaped values in errors, and prints JSON. Use --raw-json to pass or override fields that Tavily adds after this skill version.

Common Commands

# Check local auth configuration; add --execute to call GET /usage.
python3 scripts/tavily_api.py usage

# Search with source filtering.
python3 scripts/tavily_api.py search --query "RAG evaluation best practices" --include-domains arxiv.org,openai.com --include-answer advanced --execute

# Extract known pages.
python3 scripts/tavily_api.py extract --url "https://docs.tavily.com/" --format markdown --execute

# Discover URLs before extracting.
python3 scripts/tavily_api.py map --url "https://docs.tavily.com" --instructions "Find API reference pages" --execute

# Crawl a site section with a limit.
python3 scripts/tavily_api.py crawl --url "https://docs.tavily.com" --instructions "API docs only" --limit 20 --execute

# Start and poll a research task.
python3 scripts/tavily_api.py research --query "What changed in AI search APIs this month?" --model auto --execute
python3 scripts/tavily_api.py research-status "REQUEST_ID_FROM_CREATE_RESPONSE" --execute

References

Read references/api-notes.md for endpoint and option notes copied from Tavily's public API docs.
Read references/authentication.md before asking the user for, storing, or exporting credentials.