ordo-build-pipeline

Sequences six gated build steps for autonomously constructing Ordo applications.

Updated Aug 10, 2026
One-click install
npx skills add https://github.com/Rekonquest/ordo-pro --skill ordo-build-pipeline-rekonquest
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ordo-build-pipeline
Source: https://github.com/Rekonquest/ordo-pro/tree/main/default-user-files/skills/ordo-build-pipeline
Command: npx skills add https://github.com/Rekonquest/ordo-pro --skill ordo-build-pipeline-rekonquest

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Autonomous multi-step app builds drift: coders skip verification gates, procedural instructions pile up in context, and build state is lost across sessions. This Skill gives the planner a fixed sequencing spec so an Ordo app is built step by step with enforced gates and durable state. ## Core Features & Use Cases - Six gated build steps: intake, blueprint, crate build, crate couple, test, and launch proof, each with an entry condition and an exit gate that must emit a GateResult Pass before advancing. - Build ledger in persistent memory: stores requirements, versioned blueprints, per-crate and per-wiring status, deferred-debt and COUPLE-marker lists, retry history, and launch proof so builds survive session boundaries. - Skill eviction rule: drops a completed step's procedural text from context only at a passed gate, keeping context flat while outputs persist in the ledger. - Use Case: A planner agent resuming an interrupted Ordo build reads the ledger, confirms the last GateResult Pass, evicts the finished step's skill, and releases exactly the next step skill to the coder. ## Quick Start Use the ordo-build-pipeline skill to plan and advance the next gated step of my Ordo app build.

Frequently Asked Questions about ordo-build-pipeline

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I orchestrate a multi-step autonomous app build with an AI planner?▼

Define a fixed sequence of gated steps where the planner releases one step skill at a time to the coder. Each step ends with a deterministic gate classifier emitting a GateResult, and the planner advances only on a Pass, never on the coder's claim of done.

How do I keep AI agent context from growing during long builds?▼

Evict each step's procedural skill text from context when its gate passes, while writing the step's output to a persistent ledger. The next step reads only its projected ledger slice, keeping context roughly flat per step.

How do I persist build state across agent sessions?▼

Use a build ledger in persistent memory keyed to the project identity. It stores requirements, versioned blueprints, per-crate and per-wiring status, deferred-debt lists, retry history, and launch proof so the build resumes after any session boundary.

What happens when a build gate fails in this pipeline?▼

A GateResult Fail routes to the ordo-error-router and never advances the build. If the autonomous_correction flag is false, the failure hard-halts and surfaces the trace; if true, the router's bounded autonomous-correction branch handles the retry.

When should the planner not evict a step skill?▼

Never evict mid-step or mid-retry; eviction happens only at a passed gate. The error router's Debugger is exempt because it checks retry diffs against the current step's gate, so that skill stays present through every retry.