Can Claude Code, Cursor, or Codex write a Sorsa API integration for me?
Yes. Paste our docs URL into your assistant and it writes the client, the requests, and the response parsing for you. Most integrations are a single prompt.
Paste one docs URL into any LLM and ship your X integration in a single prompt. Built for AI coding assistants and for autonomous agents that need live X data. Your first 100 requests are free: no card, no expiry.
Paste the prompt, paste the docs URL, get working code. Here is the exact prompt for each tool.
Full project integration from the terminal.
Build a Python script that pulls every tweet from @naval and saves them to CSV. Use the Sorsa API, docs: https://api.sorsa.io/v3/static/sorsa_api_v3_documentation_en.txt
Generate functions and modules.
Write a function that takes a list of usernames and returns their follower counts with the Sorsa API. Docs: https://api.sorsa.io/v3/static/sorsa_api_v3_documentation_en.txt
Add an API client inside your editor.
Add a Sorsa API client to this project that searches tweets by keyword and handles pagination. Docs: https://api.sorsa.io/v3/static/sorsa_api_v3_documentation_en.txt
Inline completions.
# Sorsa API docs: https://api.sorsa.io/v3/static/sorsa_api_v3_documentation_en.txt # search tweets for "AI agents" with over 100 likes, latest first
Quick data pulls from the shell.
Use the Sorsa API to fetch the latest 100 mentions of @openai with over 50 likes. Docs: https://api.sorsa.io/v3/static/sorsa_api_v3_documentation_en.txt
One-off scripts, no setup.
Here are the Sorsa API docs: https://api.sorsa.io/v3/static/sorsa_api_v3_documentation_en.txt Write me a script to download all tweets from @sama into a CSV.
Works the same in Grok build, Antigravity, and any other LLM or coding agent. One docs URL is all it needs.
Connect any autonomous agent to a live X feed through clean tools, not glue code. The same flat 20 req/s on every plan means agents do not get throttled the moment they go autonomous, and read-only access means an agent cannot break anything.
Every endpoint maps cleanly to a tool an agent can call.
Flat 20 req/s on every plan, predictable per-request cost.
Read-only, so an agent can read X data but never post or delete.
Install the open-source Sorsa API skill once and your agent gains read-only X tools: search tweets, read profiles, pull followers and following, fetch user timelines, track mentions, and read trends. Built in the open SKILL.md format, it works in Claude Code, Cursor, and Codex CLI, with skill-based agents like OpenClaw (through ClawHub) and Hermes Agent, or in any agent that loads a skill folder. Bring your Sorsa API key. No X developer account, no posting, no write access.
Add Sorsa API as native tools in Claude, Cursor, VS Code, and any MCP client. The open-source server exposes all 40 endpoints as 40 MCP tools and installs with one npx command. Unlike the official hosted X MCP, there is no OAuth flow and no per-resource billing: one API key, flat per-request pricing, no glue code on your side.
A real round-trip. You ask, the model reads the docs, it ships this.
"Pull recent tweets about AI agents with 100+ likes and save them to CSV. Use Sorsa API."
import requests, csv resp = requests.post( "https://api.sorsa.io/v3/search-tweets", headers={"ApiKey": "YOUR_API_KEY"}, json={"query": "AI agents min_faves:100", "order": "latest"}, ) tweets = resp.json()["tweets"] with open("tweets.csv", "w", newline="") as f: writer = csv.writer(f) writer.writerow(["author", "likes", "text"]) for t in tweets: writer.writerow([t["user"]["username"], t["likes_count"], t["full_text"]])
That is the whole integration. Pass next_cursor to page past the first 20 results. Every tweet already includes the full author profile at no extra cost.
The docs are built so an LLM can read them once and produce correct, working calls without guessing.
The model reads every endpoint, parameter, and response shape from a single URL.
A native fit for Claude tool use and OpenAI function calling.
Every response has the same shape, so generated code parses it on the first try.
The cost of the code your assistant writes stays predictable, with no per-resource math.
The model's prior knowledge transfers, so it writes correct calls faster.
An assistant or agent cannot post, delete, or take any destructive action.
What an assistant or agent can actually do with X data on its own, and what it can do once you point it at Sorsa API.
| Capability | Without Sorsa API | With Sorsa API |
|---|---|---|
| Search X platform-wide | Cannot, only reads a URL you paste in | Searches millions by keyword, operator, date, and engagement |
| Read profiles and the follower graph | Limited to one pasted page | Full profile, followers, and following by handle or ID |
| Real-time data | Stale or blocked | Live tweets, metrics, and trends |
| Build a working integration | Guesses an undocumented or OAuth-gated API and stalls | Reads one docs URL and ships code |
| Run autonomously at scale | Rate-limited and brittle | 20 req/s on every plan, flat per-request cost |
Point your assistant at the docs and describe the outcome. Each of these is one prompt away.
Sentiment pipeline over any keyword or cashtag.
See howExport a follower or following list to CSV.
See howMonitor mentions of your brand in real time.
See howTrack a competitor's timeline and engagement.
See howPull trending topics for any location.
See howBuild a clean dataset for model training.
See howOur documentation is a single self-contained, machine-readable file. Drop the URL into Claude Code, Codex, Cursor, ChatGPT, Gemini, or Grok and the model learns every endpoint, parameter, and response shape on the spot, then writes your integration end to end. No SDK to learn, no wrapper to write.
The model reads everything from a single URL, no crawling 30 doc pages.
Structured for machine reading and native function calling.
Every response has the same predictable shape, so generated code just works.
Yes. Paste our docs URL into your assistant and it writes the client, the requests, and the response parsing for you. Most integrations are a single prompt.
No. For code generation, just paste the docs URL. For live agent tools you can optionally install the MCP server or the Sorsa API skill, but neither is required to start.
Yes. The OpenAPI 3.0 spec and clean, predictable JSON are built for both.
Claude Code, Cursor, and Codex CLI, skill-based agents like OpenClaw (through ClawHub) and Hermes Agent, plus any agent that reads the open SKILL.md format. You bring your Sorsa API key.
No. Sorsa API is read-only by design. That keeps autonomous agents safe and the service stable: an agent can read X data but never post, delete, or follow.
Flat pricing. 1 request equals 1 call on every endpoint and every plan, with a flat 20 req/s so the agent is not throttled.
Yes. Every new account gets 100 free requests: one-time, no card required, they never expire, and they cover all 40 endpoints. That is enough to test generated code, the MCP server, or the skill end to end before picking a plan.
Yes to both. The OpenAPI 3.0 documentation lives at one URL, and the open-source Sorsa API MCP server exposes all 40 endpoints as native tools in any MCP client. It installs with one npx command and runs on a single API key, with no OAuth setup.
Public X data: tweet and user search, profiles, followers and following, timelines, comments, quotes, retweeters, mentions, trends, and action verification.
Drop the URL into your assistant, grab a key, and your X integration is done. Your first 100 requests are free: no card, no expiry.