Tools and keys overview
Every free service Larvabot can use, grouped by job, in the order it tries them, with a link to set up each key.
Every job in Larvabot has a list of services it can use, tried top to bottom. A service is skipped when it isn't set up, has used its free quota, or is having a bad day, so adding more keys makes Larvabot more reliable, never more expensive.
AI
Drafting order. Scoring swaps Workers AI ahead of OpenRouter, because it's quicker for short answers.
- 1Google GeminikeyFree tier with daily limits (no charge unless billing is enabled)
- 2GroqkeyFree tier, rate-limited, no card
- 3OpenRouter (:free models)keyFree models, daily request limit
- 4Cloudflare Workers AIkey10,000 neurons / day, no card
- 5CerebraskeyFree tier, daily token limit
- 6MistralkeyFree 'Experiment' plan
- 7Custom OpenAI-compatible (e.g. Ollama)keyYour own hardware
Web search
Monthly quotas first, your own SearXNG next, one-time credits after, keyless last.
- 1Tavilykey1,000 credits / month
- 2Linkupkey$20 credit / month (~4,000 searches), no card
- 3Exakey$10 credit / month (~1,400 searches), no card
- 4Brave Search APIkeyMonthly free credit (card required)
- 5SearXNG (your server)keyUnlimited on your own server
- 6Larvabot search serverkeyShared server, 1,500 searches / month each
- 7Serper (Google results)key2,500 credits, one-time
- 8Jina SearchkeyFree tokens per key, one-time
- 9DuckDuckGo HTMLno keyNo key; often rate-limited
Reading pages
No keys needed.
- 1Direct fetch + HTML parseno keyUnlimited
- 2Jina Readerno key~20 req/min without key
Finding contacts
The page itself first. Paid-by-credit finders only when it shows nothing.
- 1Page + contact-page scrapingno keyUnlimited
- 2MX check (DNS-over-HTTPS)no keyUnlimited
- 3Prospeokey100 credits / month; no charge when nothing is found
- 4Tombakey25 searches / month
- 5Hunter.iokey50 searches / month
Several routes in, because Reddit blocks many servers.
- 1F5Bot email alerts (via IMAP)keyUnlimited, real-time
- 2Reddit API (legacy app OAuth)key100 requests / min
- 3Reddit via Larvabot's relaykeyNo key; runs on Larvabot's search server
- 4Reddit public JSONno keyNo key; often blocked from cloud IPs
- 5Reddit RSSno keyNo key; often blocked from cloud IPs
- 6Web search (site:reddit.com)no keyUses the search chain's quota
Other conversations
Which ones run is picked per project by the site analysis.
- 1no key
- 2Stack Exchange APIno key300 requests/day without a key, 10,000 with a free key
- 3YouTube comments (Data API)key10,000 units/day (a search costs 100, comments 1 per video)
- 4GitHub issues & discussionsno key10 searches/min without a token, 30/min with one (discussions need a token)
- 5Bluesky searchkeyFree with an app password
- 6Lemmy (federated communities)no keyPublic search, no key
- 7Discourse forumsno keyPublic search on each forum
- 8Web discussions (your search keys)no keyUses the search chain; your SearXNG first
Sending & replies
Your SMTP for sending, your IMAP inbox for everything coming back.
- 1SMTP (e.g. Brevo free)keyBrevo: 300 emails / day
- 2key