Leverage guide¶

What problem this solves¶

Without a portable package, every agent rediscovers FOIA/ETL rules from chat history. This repo gives any agent the same architecture, policy, playbooks, and MVP gate in one place.

What OKF is¶

Open Knowledge Format (OKF) v0.1 is an LLM wiki:

Markdown files with YAML frontmatter (type required on concept files)
okf/index.md for progressive disclosure
okf/log.md for changelog

Validate: python3 scripts/okf_validate.py okf --strict

Three layers (+ infra)¶

Layer	Question it answers	This repo
OKF	What do we know?	`okf/`
Skills	How should I behave right now?	`skills/*/SKILL.md`
Harness	How do we prove it works?	`harness/e2e.sh`
Code + infra	What runs?	`src/`, `infra/`

Rule of thumb: durable facts → OKF; triggers and hard rules → skills; proof → harness; execution → Python/Terraform.

Day-to-day agent use¶

Point agent at this repo
  → AGENTS.md
  → okf/index.md
  → skill for the task
  → linked playbooks
  → run ./harness/e2e.sh before share/deploy claims

Adopter ladder¶

Level	Action
0	`./harness/e2e.sh` — prove MVP
1	Run local MVP
2	Extend new source
3	SCALING.md — local → GCP ladder
4	FINAL-REVIEW.md pre-scale checklist + production HITL

Repo is public (Apache-2.0). Share PDFs from docs/share/ for slides; link the repo for proof. See QA before share.