Features
Everything your
library can do.
A portable body of work — and the power to put it to work.
Factor.In keeps your notes, prompts and skills as plain-text files you own — then runs them against real work: a spreadsheet, a codebase, a manuscript, with the model you choose.
Here's what comes with it.
Highlights
Run it, watch it work
Big tasks, in one shot.
Write a task, point it at your files, and Factor.In chains your documents, prompts and skills into an ad hoc workflow. It runs a full agentic loop — many turns of real work — with a live feed you can watch or stop, and comes back done.
A full audit trail
Every asset, accounted for.
Built-in versioning records how each file in your project came to be — what ran, what it changed, what it cost, down to the token. Not a black box: a legible history of your whole project.
Approve, reject, roll back
Nothing changes without your say-so.
Every run returns a proposed new state for you to read before you commit. Accept it, reject it, or roll back to any earlier point in time. The canonical version is always the one you approved.
Everything else
Simplicity
Only two ideas to learn.
The whole system is prompts and skills — plain-language instructions, and the reusable know-how to carry them out. You can safely ignore the rest of the acronym soup.
Focus
Documents, not chat.
Turn-by-turn chat is low-bandwidth, and buries your thinking in a scroll you'll never reread. Documents keep your intent legible, reorderable and reusable — so a good idea earns its keep more than once.
Portability
Plain text. Profoundly yours.
Everything is markdown you can read without us — no proprietary format, no lock-in, no "memory" that quietly becomes the vendor's. Export the whole library in a single click and take it anywhere.
Compounding
It grows and travels with you.
Your library is your accrued, AI-native expertise — carried from project to project, not copied over, gone stale or forgotten. What you learn once, you keep.
Shareable skills
Expertise you can hand to someone.
A skill is just a file — or a folder of them — that explains, in plain language, how to do something well. Pull them from GitHub or a teammate, tweak them to taste, and share your own back.
Flexibility
The right model for the job.
Mix and match providers and models — the best one for a task, or for one part of a task. Bring your own keys or use ours. You're never wedded to a single vendor's roadmap.
Off-device sandbox
Runs at arm's length.
Executions happen off your device, in a secure sandbox — never on your machine with your keys and unrelated files exposed. The work happens elsewhere; only the result comes home.
Context engineering
A clean desk, every run.
Each execution gets its own fresh context window, anchored in the files you chose and nothing else. No bloated history, no context rot — just the right material, every time.