Bansat (بنسټ)

Bansat is Pashto for foundation: lay it before you build. Before any production code, your AI agent writes the plan a senior team would: product, architecture, security, privacy law, accessibility, every UI state and how the project will run. Then it stops and waits for a person to approve it.

Plan before code

One blueprint holds the PRD, TRD, app flow, UI/UX brief, backend schema, implementation plan and decision records, so they can’t contradict each other.

Gates a person signs

Five gates, from idea to the 30-day review. Only a person signs one, and an optional Claude Code hook makes the first a hard stop.

Privacy and security built in

UK and EU GDPR, PECR, the EU AI Act and Nigeria’s NDPA, checked against official sources, plus 17 launch traps to clear before real data.

Fits how you work

A picker recommends Scrum, Kanban, stage gates, Shape Up, a solo loop or a scaled method, and shows where every gate falls.

Install

For Claude Code, clone the latest release into your skills folder and type /bansat:

git clone --branch v2.4.0 --depth 1 https://github.com/bewithkhan/bansat ~/.claude/skills/bansat

In the Claude apps, add it as a skill. Any other agent can read the framework as Markdown.

Read the framework

How to use itTracks, depth by risk tier, the gates, and where the usual planning documents live.

Part 0: intake and riskThe screening and delivery questions every project starts with.

Breach casebook20 documented incidents from 2025 and 2026, each mapped to the checks that catch its class of failure.

Delivery methodsWhere each gate falls in Scrum, Kanban, stage gates, Shape Up, a solo loop and scaled methods.

Tested examplesThe framework's code, run in CI: the schema, the access rules, the consent loader and the privacy test.

Not legal advice. The legal sections were checked against official sources with AI assistance on 4 November 2026, and no lawyer has reviewed them yet. Text and templates are CC BY 4.0, and code samples are MIT. Created by Muhammad Tufail Khan.