Launch sentiment sweep with Veezee
Turn a launch or announcement name into a one-time readout of how Reddit and X reacted in a given window: how much volume, what people actually said, and who the loudest voices were. Reddit and X only; this skill covers no other platform. Veezee returns raw posts, comments, and tweets, never a sentiment score: you, the agent, read the returned text and judge tone yourself.
Setup (one free key, no signup)
Every call needs an API key. Minting one is free and human-free: no signup, no card. A call without a key fails with KEY_REQUIRED (401) carrying mint_url; the fix is the free mint below, never a payment. Two equivalent surfaces; pick whichever your environment has:
- MCP: add
https://mcp.veezee.io/reddit and https://mcp.veezee.io/x (streamable-http, one platform each), or https://mcp.veezee.io/all for both plus LinkedIn in one mount. Hosts that support MCP authorization (Claude Code, claude.ai) open a Veezee sign-in on connect (email code, no password): that is the whole auth step. Other hosts: mint a key with POST https://api.veezee.io/v1/keys/mint (empty body; the key is shown once) and put it in the connection's Authorization: Bearer header.
- SDK:
import { VeezeeClient } from "@veezee/sdk"; const client = new VeezeeClient(); await client.mint(); mints and stores the free key the first time and reuses it on later runs. Reddit methods live on client.reddit.search/getSubreddit/getSubredditPosts/getUser/getPost/resolveUrl; X methods follow the same per-platform pattern (client.x.search, client.x.getProfile, client.x.getTweets, client.x.getTweet, client.x.resolveUrl). client.resolveUrl and client.getUsage are also available top-level. The veezee CLI (npx @veezee/sdk init) mints and stores the same key.
A fresh vz_trial_ key runs under a free per-IP daily budget of 200 credits shared across platforms, cached data only, first page only on any cursor sequence. That covers a small test sweep; a real two-platform sweep needs purchased credits. When the budget or a trial cap runs out (TRIAL_CAP_EXCEEDED, INSUFFICIENT_CREDITS, BUDGET_EXHAUSTED), the error carries upgrade_url and a machine-readable offer: hand that link to your human; purchases credit the same key directly, nothing to reconfigure.
The sweep
- Check the budget first:
get_usage is free, exempt from the rate limit, and works the same on a trial or paid key.
- Pick the window.
reddit_search accepts range (past_hour..all_time) for type: "posts" only; comment search has no server-side time filter, so for type: "comments" fetch with sort: "new" and filter results yourself by created_at. x_search has no date parameter either: put since:/until: operators directly in the query string.
- Sweep Reddit:
reddit_search with type: "posts", the launch name as query, and range set to the window, for what people are posting; a second reddit_search with type: "comments", sort: "new" for reactions buried inside other discussions, the more unique signal. Decompose a broad or generic launch name into narrower queries (an exact phrase, the product name plus a qualifier) if one query looks capped near its result depth.
- Sweep X:
x_search with type: "recent" for the chronological deep sweep (follow cursor as far as the volume warrants; on the free tier only the first page is served, deeper pages need purchased credits); type: "popular" for the highest-engagement takes, which are usually worth quoting even at low volume.
- Drill into the threads and tweets that carry the most signal (most comments, most engagement, or clearly representative of a theme). Batch-fetch Reddit bodies with
reddit_get_post (up to 100 ids per call, first 10 included in the base price); reserve detail: "full" (+4 credits, one id only) for the one or two threads worth reading end to end. For a specific high-engagement tweet, x_get_tweet gets full metrics.
- For accounts that matter (an account whose take is driving the conversation, or whose follower count makes them worth naming),
x_get_profile for bio and follower count. Do not profile every author; that is not what the budget is for.
- Read the returned text yourself and sort it into positive, negative, and neutral. Veezee has no sentiment-scoring endpoint; the API returns raw text and engagement counts, and judging tone is your job as the agent, not a field in the response.
Rules that save credits and errors
- Default freshness is cached (
recent, usually refreshing every 10-20 minutes) and free. Launch-day windows often justify freshness: "realtime" (+2 credits per call) to catch reactions from the last few minutes rather than the last cache cycle, but it needs a PAID key: trial keys are cached-only and reject realtime with TRIAL_CAP_EXCEEDED.
- Set
max_credits on each call; a call whose quote exceeds it is rejected with nothing charged, so a sweep can keep going instead of overspending on one query.
- On KEY_REQUIRED, mint the free key and retry the same call; this one is never a payment problem. On TRIAL_CAP_EXCEEDED, INSUFFICIENT_CREDITS, or BUDGET_EXHAUSTED, stop and hand the error's
upgrade_url to your human; purchases credit the account directly and the same key keeps working afterward. On RATE_LIMITED (429), wait retry_after_seconds and continue; it clears on its own and payment only raises the limit.
- A 403 NOT_ENTITLED means the key's account is not enabled for that platform; grants are explicit, write to hello@veezee.io to enable more.
- This is a one-shot sweep of a fixed window, run once and reported, not a running loop. For ongoing, deduped-across-polls coverage of a topic or brand over time, use the reddit-monitoring skill instead.
Output
A structured readout the human can act on without re-reading the raw data. Use this shape:
# reaction sweep ()
Volume: Reddit posts, Reddit comments, tweets (list each query and the window it covered)
Positive / Negative / Neutral: 2-3 representative quotes per bucket, each with author and link
(`permalink` for Reddit, `url` for X)
Themes: 3-5 recurring topics, one line each
Notable accounts: name, follower count, stance, why they matter to this launch
Credits spent:
Quotes must be verbatim from returned text with real links; never paraphrase into the quote slot. Report credits spent so the human can budget the next sweep.