fde-engineeringlisted

# @fde-engineering ## Purpose FDEOS is an operating system for one engineer running an engagement end to end. This skill is the **execution layer**: how you build, clean up, review, and verify on the client site without a team behind you. Engagement judgment lives in land, discover, and rescue. **This skill is how you ship at staff-engineer rigor once scope is clear.** Talk like a tech lead pair-programming — status in plain English ("here's what I changed, what I ran, what's still risky"), not a compliance checklist. ## Token efficiency Load `context.md`, `decisions.md`, and `success.md`. Load `terrain.md` only when touching code. Load `trust-profile.md` when regulated data or AI policy applies. ## When to use - You are implementing a feature, fix, or integration on a client codebase. - You need a repeatable loop: small change → verify → cleanup → review → ship. - You are solo and cannot afford thousand-line PRs or vibe-coded mess. Do not use for a one-line typo in your own repo with no engagement context. ## The FDE agentic loop Always in this order: 1. **Confirm scope in writing.** Read `success.md` and the relevant slice in `decisions.md`. If the task is not in the plan, run `@fde-plan` or name scope creep before coding. 2. **Keep the unit small.** One feature, one fix, one reviewable PR. If the plan item is bigger than ~90 minutes of focused work, split it in `decisions.md` before starting. 3. **Search before creating.** Read existing code in the area you w