Agents that operate on social platforms need more than a scraper and a session cookie. They need structured, composable skills — ones that hold up under load, respect platform boundaries, and don’t expose credentials every time a new deployment spins up. That’s what AIsa Twitter Skill for OpenClaw is designed to do.
Two versions, two distinct architectures. Here’s how we got here.
v1 — Read the room
The first version of the AIsa Twitter Skill was deliberately scoped. Routed through AIsa’s unified API at api.aisa.one/apis/v1/twitter/, it gave OpenClaw agents a clean interface into X’s read surface: timeline queries, keyword search, profile lookups, thread traversal. The same API key your agent already uses for other providers works here — no additional credential management.
Write operations were technically present — agents could post tweets, threads, and replies — but the mechanism was a login cookie: a cloned browser session handed to the agent. It worked inconsistently. Cookies expired without warning, X’s bot detection flagged the traffic, and posts would silently fail mid-run. In practice, most teams used v1 for read-only workflows and avoided relying on the write surface for anything that mattered.
v1 — Deprecated
- Timeline queries and thread traversal
- Keyword and hashtag search
- User profile and follower data
- Write ops via login cookie — unreliable
- Prone to bot detection & silent failures
v2 — Current
- All v1 read & query capabilities (timelines, search, profiles, threads)
- Post tweets and threads
- Reply to conversations
- OAuth 2.0 — no cookie exposure
- Engagement actions (coming soon)
v2 — The write surface, done right
The major architectural shift in v2 is the authentication model. Rather than inheriting the cookie-based limitations described above, v2 replaces that approach entirely with OAuth 2.0. Your account credentials never leave your hands. The agent receives a scoped access token authorised specifically for the actions it needs to perform — posting, threading, replying. Token rotation is handled automatically. There’s nothing to leak.
Why it matters: OAuth isn’t just a security upgrade — it’s a reliability one. Requests arriving via a properly authorised OAuth flow look like normal API traffic to X’s systems. No heuristics trip. No sudden throttling. Your agent posts when it’s supposed to post.
Write operations in v2
The current write surface covers the three core publishing primitives:
| Operation | v1 | v2 |
|---|---|---|
| Read timelines & search | ✓ | ✓ |
| Post a tweet | ~ | ✓ |
| Post a thread | ~ | ✓ |
| Reply to a tweet | ~ | ✓ |
| Engagement actions (likes, retweets, follows) | — | Coming Soon |
~ available via login cookie — inconsistent, bot-detection prone
How it fits into OpenClaw
The AIsa Twitter Skill is one of the most sought-after OpenClaw skill, published on ClawHub in late January and already surpassing 10K downloads. It’s fully composable with MarketPulse and the Tavily search and Perplexity search skills — all of which leverage AIsa’s unified API. You configure it once in your skill manifest and any agent in your deployment can invoke it by name.
The OAuth token is stored at the platform level — not inside the agent, not in your prompt. Agents call the skill, the skill holds the auth context, AIsa handles the routing. The separation is intentional: your agents can be stateless and your credentials stay scoped.
A typical workflow looks like an intelligence + creation + publishing loop: the agent monitors mentions and keyword trends via the read operations, drafts tweets or threads, and eventually publishes through the write operations either instantly or on a confirmed schedule. Each step is a separate skill invocation, composable with any other skill in the registry.
Getting started
The simplest way to get the latest AIsa Twitter Skill into your agent: feed the GitHub URL directly to your OpenClaw agent.
The agent reads the skill definition straight from the repo — always current, no manual version pinning. Questions land in the AIsa Discord.
Ready to give your agent a voice on X? Install the AIsa Twitter Skill from ClawHub or point your OpenClaw agent directly at the GitHub repo to get started.