Crawls a site and scores how citable it is to AI engines, then returns a prioritized fix list.
Why it matters
Know exactly why ChatGPT, Perplexity, and Google AI Overviews skip your pages, and what to fix first to start earning citations.
26 production-ready recipes
AI Workflow Recipe Library.
Real, agentic AI workflows you can ship: not toy demos. Each recipe is the architecture, the exact stack, and the guardrails that keep it alive in production. Browse the build, then take it to your own use case.
- Recipes
- 26
- Categories
- 8
- Skill range
- Beginner → Advanced
The anatomy
Every recipe is the same five-stage pipeline.
Different sources, different models, different destinations. The shape underneath is always this: ingest the world, reason over it cheaply, validate hard, and deliver where work happens. Two rails run the length of it.
- Trigger A schedule, a webhook, or a user query kicks off the run. Cron · Events · API
- Ingest Fetch and render the messy source into clean, typed text. Crawl4AI · RSS · SERP
- Reason A cost-routed model classifies, extracts, scores, or writes. OpenRouter · LLMs
- Validate Schema checks, grounding, and guardrails before anything ships. Schemas · Human review
- Deliver Push the result where the work actually happens. Telegram · Sheets · CRM
Memory Hermes, vectors, and key-value stores give the workflow state across runs.
Observability & cost control Structured logs, budget caps, and model fallbacks keep it cheap and debuggable.
Category
Difficulty
No recipes match those filters yet.
Compares your coverage to the entities and questions competitors rank for, then outputs prioritized briefs.
Why it matters
Stop guessing what to write. Get a ranked, evidence-backed list of the gaps worth closing, grounded in what is actually on the SERP.
Finds the published pages aging out of AI Overviews, ranks them by decay, and queues a grounded refresh for each before the citations slip.
Why it matters
Freshness is a top citation signal, so pages quietly fall out of AI answers as they age. This keeps a steady queue of the highest-impact updates instead of a once-a-year content audit.
Builds a resolved entity knowledge base for your brand, links it to public knowledge graphs, and emits the schema that makes AI engines understand who you are.
Why it matters
AI engines cite sources they can model as entities. A clean, linked entity base makes your people, products, and concepts legible to them, which is the groundwork for being cited at all.
Runs every draft Etsy listing through a QA pass, title, tags, images, SEO, and policy, and returns a pass or fix-it checklist before you hit publish.
Why it matters
A listing published with weak tags or a thin description leaks sales quietly for months. This catches the fixable problems in seconds, so every launch goes out at full strength.
Watches a list of ASINs and alerts you the moment a real price drop clears your buy threshold.
Why it matters
Stop refreshing product pages. Buy at the genuine low with full context, current price versus thirty-day floor, the instant it happens.
Researches an Etsy niche end to end and returns a ranked shortlist of product opportunities with rationale.
Why it matters
Replace a week of manual market research with a grounded shortlist that scores demand against competition, so you build what is actually underserved.
A resilient, LLM-assisted pipeline that turns messy retail catalog pages into clean structured product data at scale.
Why it matters
Get a dependable product dataset that survives site redesigns, the failure point that breaks most home-grown scrapers within a month.
Routes each request to the cheapest model that can actually handle it, with automatic fallback and budget caps.
Why it matters
Cut LLM spend by a large margin without losing quality. Stop paying frontier prices for trivial extract-and-classify calls.
A reference pattern for an assistant that remembers across sessions while keeping the context window small and on-topic.
Why it matters
Ship an assistant that feels like it knows the user without ballooning prompt cost. Separating working from long-term memory is the difference between a demo and a product.
Gives every agent its own Durable Object as a strongly consistent memory store at the edge, with SQLite state, alarm-driven decay, and hibernating WebSockets.
Why it matters
Per-user agent memory usually means a database round trip and a consistency headache. A Durable Object collapses state, compute, and locality into one addressable object, so each agent's memory is fast, consistent, and isolated by construction.
Wires a coding agent to pull current, version-pinned library docs through Context7 before it writes a line against an SDK.
Why it matters
Kill the most common failure of coding agents: confidently calling APIs that changed or never existed. Ground generation in real docs and the diff actually compiles.
Spins up a per-branch preview deploy, smoke-tests it on the live URL, and posts a pass or fail verdict back to the pull request.
Why it matters
Catch broken builds and regressions before they reach main. Every branch gets a live URL and an automated verdict instead of waiting on someone to click around.
Reviews each pull request, pushes low-risk fixes as commits, and loops until the checks pass or it hands back to a human.
Why it matters
Cut review latency from days to minutes on routine PRs. Reviewers spend their attention on design, not on lint, missing tests, and obvious bugs.
Fans out a fleet of agents across a codebase to find dead code, duplication, and needless complexity, then verifies each fix in isolation before it ships.
Why it matters
Pay down tech debt at a pace one engineer cannot match. Smaller, simpler code is cheaper to run, faster to change, and easier for the next agent to reason about.
Turns the raw transcripts from an Omi wearable into structured action items, decisions, and follow-ups routed to where you actually work.
Why it matters
Stop losing the good idea you had on a walk. Every captured conversation becomes searchable notes and tracked tasks, with zero manual journaling.
Turns a recorded meeting into attributed decisions and assigned action items, syncs them to your tracker, and chases the ones that go unfinished.
Why it matters
Most meeting notes die in a doc nobody reopens. This assigns every action to a real owner, files it where work actually happens, and follows up so commitments close instead of evaporating.
A self-hosted voice front end to a Hermes agent with streaming speech, barge-in, and the full tool and memory stack behind it.
Why it matters
Give customers or staff a natural voice line into your agent, on the web or over the phone, without renting a black-box platform.
Collapses a wall of industry RSS feeds into one ranked, summarized Telegram digest each morning.
Why it matters
Read five items that matter instead of scanning two hundred. Stay ahead of your space in five minutes a day, fully hands-off.
Asks the answer engines your money questions on a schedule and alerts you when your citations appear, vanish, or get replaced by a competitor.
Why it matters
GEO has no rank tracker yet, so build your own. See exactly when ChatGPT, Perplexity, or Google AI Overviews start or stop citing you, before the traffic moves.
Watches competitor pages for meaningful change, pricing, positioning, launches, hiring, and tells you what moved and why it matters, filtering out the noise.
Why it matters
Know your competitor moved before your prospect does. Turn scattered page-watching into one signal feed of real strategic changes, without a person diffing screenshots.
Builds a baseline of your LLM spend and pages you the moment a key, app, or route burns money abnormally, a runaway loop, a leaked key, a model price jump.
Why it matters
LLM bills do not spike politely; one looping agent or a leaked key can run thousands overnight. This catches the anomaly in minutes, attributes it, and can throttle before the invoice does the talking.
Researches a shortlist, drafts genuinely personalized first-touch messages grounded in real profile facts, and queues them for human review.
Why it matters
Scale outreach without sounding like a mail merge. Personalization that is actually true lifts reply rates and protects the brand.
Captures every inbound recruiter message and job lead, scores it against your criteria, tracks it through a pipeline, and drafts the follow-up before it goes cold.
Why it matters
Good opportunities die from slow or forgotten replies, not from lack of interest. This keeps a scored pipeline of every inbound role and nudges you to respond while the conversation is still warm.
Triages your inbox and calendar each morning into one brief: what needs a reply, what to decline, and what to prep for.
Why it matters
Reclaim the first hour of the day. Instead of a sixty-message inbox and a wall of invites, you get a ranked brief with drafts ready for one click.
Continuously syncs your notes, highlights, and saved links into one entity-resolved knowledge graph your agents can actually query.
Why it matters
Your knowledge is scattered across ten apps. Unify it into a single graph every agent shares, so your assistant answers from what you already know, not the open web.
From recipe to production
A recipe is the map. I build the road.
Every workflow here is one I can stand up on your stack, wired to your tools, with the guardrails that keep it running when you are not watching. Start with a free 30-minute call: bring the problem, and we will scope the build.