Project Delivery Engine helps AI agents reliably manage long-running projects with persistent state, structured handoffs, and independent review.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install project-delivery-engine
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install project-delivery-engine using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Project Delivery Engine?
Project Delivery Engine is an Openclaw Skills workflow for keeping AI-driven projects on track over days, weeks, or months without losing context, progress, or delivery quality.
It separates project state, evidence, collaboration, and governance so the main agent can continue working steadily while spawning parallel agents, using isolated worktrees, and handing important outcomes to independent review before final delivery. The result is a durable project-management layer for Openclaw Skills that prevents drift, preserves continuity, and makes complex multi-step work safer to execute.
Project Delivery Engine Use Cases
- Managing a current project that must retain progress, next steps, risks, and delivery context across sessions.
- Starting a new project from an existing codebase, five-piece set, or inherited workstream.
- Continuing a paused project after switching windows, models, or returning months later.
- Organizing delivery files, status notes, and handoff materials as project scope grows.
- Running multi-agent work with spawn agent tasks, worktrees, and parallel execution.
- Requiring independent review for important outputs instead of self-approval by the same agent.
- Preparing project closure, transition notes, and clean handoff documentation.
How Project Delivery Engine Works
- The agent inspects the user’s latest instruction and the current project files to determine whether the task is tied to existing project state or is a one-off request.
- If the task depends on project state, the skill directs the agent to read
references/project-state.mdto identify the project root, five-piece set, and required indexes. - For file creation or modification, the agent consults
references/project-file-templates.mdto keep documents consistent with the project’s metadata and structure. - The agent executes straightforward mainline work directly, while collaboration, parallelization, worktrees, and confirmations are handled according to
references/collaboration.md. - When real task decomposition is needed, the agent reads
references/taskcard-evidencebundle.mdand may spawn agents or split work into isolated streams. - If governance-controlled changes are required, the agent must draft a
project-gov proposeplan first and only then apply it if the toolchain can express the change safely. - At the end of a task, the agent re-checks project state to decide whether the project should be checkpointed, documented, or cleanly handed off.
Project Delivery Engine Setup
- Select the Openclaw Skills skill for project delivery and use it only when the task depends on current project state or explicit project management needs.
- Read the project references as needed:
references/project-state.mdreferences/project-file-templates.mdreferences/collaboration.mdreferences/taskcard-evidencebundle.mdreferences/project-gov-cli.mdreferences/quickstart-faq.md
- For governed changes, always follow the propose-then-apply flow.
# Governance workflow example
project-gov propose
project-gov apply
- For machine checks, numbering, launch phrases, and governance file writes, use the project governance CLI references.
- If you are unsure how to use the skill or encounter a
project-goverror, consult the quickstart FAQ before proceeding.
Project Delivery Engine Data Schema & Taxonomy
Project Delivery Engine organizes project knowledge into a small set of durable reference files and state artifacts so Openclaw Skills can preserve continuity.
| Artifact / Reference | Purpose | Notes |
|---|---|---|
references/project-state.md |
Defines current project state, root, indexes, and continuation cues | Primary source for long-running continuity |
references/project-file-templates.md |
Standard templates for creating or repairing project files | Use before adding new docs |
references/collaboration.md |
Collaboration, parallel work, worktree, and confirmation routing | Guides multi-agent execution |
references/taskcard-evidencebundle.md |
Task breakdown and evidence packaging | Used for real decomposition and review |
references/project-gov-cli.md |
Governance CLI usage, machine checks, numbering, launch phrases, and file writes | Controls governed changes |
references/quickstart-faq.md |
First-use help and project-gov troubleshooting |
Fallback for usage questions |
Taxonomy and rules:
- Mainline execution owns final delivery and merge decisions.
- Collaboration streams, scripts, and evidence bundles cannot replace the mainline’s completion responsibility.
AGENTS.mdis limited to project identity, proprietary facts, authorization, and boundaries; it should not duplicate general skill rules.- Governance changes must be expressed as
project-gov propose→applywhenever documentation, checkpoints, or handoff writes affect governed scope. - Only the references required for the current task should be loaded, and selected files must be read fully once chosen.
Project Delivery Engine Advanced Features
- Persistent project state that survives window switches, model changes, and long gaps between sessions.
- Multi-agent support via spawn agent workflows and isolated worktrees for parallel execution.
- Independent review for critical outputs, reducing self-approval risk and improving delivery quality.
- Governance-first change management using a propose-then-apply pattern for controlled writes.
- Structured handoff support with checkpoints, delivery notes, and project closure materials.
- Minimal daily overhead with on-demand reference loading, keeping Openclaw Skills lightweight but durable.
SKILL.md
Loading
Related Openclaw Skills
OpenClaw Deal Scout
Openclaw Deal Scout automates Gmail deal triage, CRM logging, alerts, and approved replies through a local MCP gateway.
USCIS Case Status
A headless command-line skill that checks USCIS case status by receipt number using Selenium and undetected-chromedriver.
DHX Electronic Components Sourcing Agent
Openclaw Skills for DHX Tech turns BOMs, part numbers, and sourcing requests into real-time inventory checks, cross-references, and procurement-ready recommendations.
WeChat Mini Program Development Tool (Free)
Openclaw Skills for WeChat mini program development helps personal developers enforce structure, avoid common pitfalls, and optimize performance.
Dawn Memory Architecture v7.1
Openclaw Skills for Dawn Memory Architecture v7.1 creates a local, layered memory system for agents with truth hierarchy, WAL safety, and semantic search.
ai-content-creator-assistant
ai-content-creator-assistant generates multiple platform-adapted content drafts from one topic, then delivers them through a compliant, payment-verified Openclaw Skills workflow.
Featured*
