$ sudoscience

$ sudo science

# permissionless innovation — science with privileges

projects

Our research investigates human and AI experience for everything from local development to critical production systems

Webpoke

prototype

a poke with a hint, then a ranged pull

Webhooks should not deliver truth in flight. A signed, content-free poke plus a cursor-owned ranged pull turns every delivery failure into latency instead of corruption.

/projects/webpokewebpoke.sudoscience.dev

EZAF

prototype

gain control over your agentic work by separating the intent from the execution

Apps may request agentic work, but users control review, execution, routing, and result release.

/projects/ezafezaf.sudoscience.dev

DESH

prototype

the terminal, with the amnesia removed

Twenty-plus years of daily terminal use, and the verdict is in: terminals and shells both suck — not at the edges, at the defaults. Warp, Fig, and nushell each fixed a symptom; the model survived them all.

/projects/deshdesh.sudoscience.dev

Layers

prototype

user-controlled meta-apps over the web

Layers are independent apps that run on top of existing websites, with their own domains, permissions, identity, data, UI, communities, and economic models.

/projects/layerslayers.sudoscience.dev

Avoidant

prototype

structured attachments — portable artifacts with affordances

Attachments that carry data, schema, behavior, provenance, and executable handles across email, decentralized social protocols, documents, WASM, MCP (Model Context Protocol), GraphQL, and other introspectable systems.

/projects/avoidant

Contags

prototype

context tags with lifecycles, not participant-scoped streams

Apple, Google, and carriers could ship contag inboxes tomorrow as an opt-in UI feature — context lists, lifecycle chips, per-contag unread. Nothing stops SMS, RCS, MMS, and iMessage clients from adding it; `#tags` work without them.

/projects/contags

Unfiled

prototype

code is a graph, not files in folders

The current disruption is a forcing function: to delegate work you must say what you mean, declaratively and intentionally — and once you do, API boundaries move to where they should have been all along.

/projects/unfiledunfiled.sudoscience.dev

instagrate-me

prototype

machine-readable domain observability

A governed surface for seeing what machines try to discover about an organization, then publishing and monitoring the metadata they need.

/projects/instagrate-meinstagrate-me.sudoscience.dev

CloudCap

prototype

enforceable near-real-time spend limits — before the agent loop finishes

Monthly invoice alerts don't stop an agent looping right now. CloudCap puts enforceable spend limits at the proxy — trip in milliseconds when a limit spec is violated, not when the billing cycle closes.

/projects/cloudcap

Rhett

sketch

field agents on every page — rhetoric, operationalized

Page-local research agents that continuously map the living field around each investigation on this site: prior art, credible actors, felt pain, counterexamples, and translations for whoever just arrived.

/projects/rhett

method

design → implement → evaluate

We study systems as we build and use them

design

Design shapes the interaction. The sketch is a hypothesis about how a system should feel to inspect.

implement

Implementation exposes the constraints. If it can't be built, the idea wasn't finished.

evaluate

Evaluation turns use into evidence. Deployed prototypes are the lab's instruments.

clarity through multiple perspectives

We look at every problem through several lenses, focused on revealing what is essential complexity and what is accidental complexity.

the comfortable lens

How would we achieve this with the stack the team already trusts — and what does that cost?

the state-of-the-art lens

How would we achieve it (or reformulate it, or avoid it entirely) with the best available tools — and what does that cost?

the first-principles lens

How would we achieve it if nothing existed yet — and what does that cost? The gap between the three answers is the accidental complexity.

idea → product

What is the shortest path from an idea to a working product?

accidental complexity

How much of our work is the problem, and how much is the packaging? What can we do about the packaging?

zero → creator

What is the shortest path from not knowing how to program to being able to make something?

The flywheel: better tools make teaching and learning faster, which makes the next tool better. Exhaust becomes tutorials; sessions become examples; learners become teachers.

faq

the questions are real. they keep getting asked. we ask why.

“How do I sandbox a coding agent?” “Why didn't my webhook fire?” “Feature folders or layer folders?” — real questions, answered by showing why the framing can't produce a good answer, and which investigation it routes to.

limbo

the small plates menu

Not everything here has collapsed into a project yet. The uncommitted ideas live in superposition on a tasting menu, served by the house — order one enough times and it graduates.

collaboration

making ecosystems sustainable.

Collaborating with builders, labs, standards groups, institutions, and communities working near interfaces, protocols, environments, provenance, evaluation, and machine-readable surfaces.

$ mail hello@sudoscience.dev