skills.rest
ogiboy
ogiboyCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

ultrawork

Orchestrates parallel execution lanes with acceptance criteria and evidence-based verification.

Updated Feb 1, 2024
Search Tags:#workflow automation#state management#parallel execution#agent delegation#acceptance criteria#task orchestration#evidence lanes
One-click install
npx skills add https://github.com/ogiboy/portfolio-app --skill ultrawork-ogiboy
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ultrawork
Source: https://github.com/ogiboy/portfolio-app/tree/main/.codex/skills/ultrawork
Command: npx skills add https://github.com/ogiboy/portfolio-app --skill ultrawork-ogiboy

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Sequential execution of independent tasks wastes time, and unstructured delegation often produces unverified results. Ultrawork provides a disciplined parallel execution engine that grounds context first, defines pass/fail acceptance criteria before editing, and closes with evidence rather than assumptions. ## Core Features & Use Cases - Parallel Lane Execution: Runs a direct-tool lane alongside background evidence lanes for tests, docs, regression mapping, or repo analysis, firing independent agent calls simultaneously. - Deliberate Self-vs-Delegate Decisions: Classifies work by dependency shape and delegates only bounded, independent slices using tiered agent selection from references/agent-tiers.md. - Lightweight Verification Boundary: Requires build/typecheck passes, affected tests, and manual QA notes before claiming completion, without owning durable persistence or architect sign-off. - Use Case: When updating a CLI hook and its test file, keep the shared-file edits local while delegating keyword-detector regression mapping to a test-engineer lane, then verify with build and test runs. ## Quick Start Ask the assistant to run ultrawork on your set of independent tasks so it defines acceptance criteria and executes them in parallel lanes with evidence-backed verification.

Frequently Asked Questions about ultrawork

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

FAQPage Schema
How do I run multiple coding tasks in parallel with an AI agent?▼

Invoke ultrawork when tasks are independent: it classifies work by dependency shape, fires independent agent calls simultaneously, and runs a direct-tool lane alongside background evidence lanes. Shared-file or prerequisite-heavy work stays local or staged sequentially.

When should I use ultrawork instead of ultragoal or team?▼

Use ultrawork for in-session parallel execution with lightweight evidence. Choose ultragoal when you need durable goal tracking, ledger checkpoints, or resume across stories, and team when you need coordinated tmux workers with shared task state and mailbox dispatch.

How does ultrawork decide between local execution and delegation?▼

It keeps work local when the next step depends on immediate repo context, shared files, or tight iteration. It delegates only bounded, independent slices that materially improve throughput, always passing the model parameter explicitly and consulting references/agent-tiers.md for tier selection.

Does ultrawork guarantee verified completion or persistent state?▼

No. Ultrawork owns only lightweight verification: build/typecheck passes, affected tests, and manual QA notes. Durable persistence, architect verification, and the full verified-completion promise belong to ultragoal, team, or the explicitly requested ralph fallback.

What evidence is required before ultrawork reports completion?▼

Completion requires the acceptance criteria defined before execution: relevant build or typecheck passes, affected tests passing, manual QA notes for user-visible behavior, and no new errors introduced. Claims without verification output violate the protocol.

Related Skills

Explore More Software Engineering Skills→
HaydernCenterpointHaydernCenterpoint

deep-interview

Clarify vague project ideas through a Socratic interview into a testable spec.

Community
Advanced
Leap0920Leap0920
1

ultragoal

Persist multi-step goal progress across Claude sessions with checkpointed artifacts.

Community
Advanced
KirillKlemKirillKlem
1

verify-external-api-docs

Verify external API behavior against official documentation.

Community
Intermediate
lpjhelderlpjhelder

team

Coordinate multiple workers on a shared task list via tmux-based orchestration.

Community
Advanced