A plan-first framework for managing non-trivial AI agent tasks through a rigorous lifecycle of tracking, context packaging, and verifiable execution.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install workflow-execution
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 workflow-execution 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 Workflow Execution?
The Workflow Execution skill provides a formalized methodology for handling complex, multi-step engineering tasks that go beyond simple chat commands. By prioritizing planning and structured documentation over transient chat history, this skill ensures that AI agents have a persistent source of truth for architecture decisions, risk assessments, and verification criteria. Integrating this into your collection of Openclaw Skills enables high-reliability workflows where agents can be spawned, handed off, and audited with complete technical transparency.
This approach shifts the focus from optimism to explicit proof, requiring agents to document failure modes and done criteria before acting. It is particularly valuable for teams using autonomous agents for migrations, security-sensitive edits, or iterative bug fixing where context loss or session compaction could otherwise derail the implementation process.
Workflow Execution Use Cases
- Tasks involving 3 or more distinct implementation or architectural steps.
- Projects requiring changes to shared contracts, interfaces, or core data flows.
- High-risk operations including security updates, authentication changes, or production configuration migrations.
- Complex work sessions that need to be handed off from a planning agent to one or more specialized executing agents.
How Workflow Execution Works
- Phase 1: Plan - Define the goal, checkable done criteria, constraints, and failure modes before any code is modified.
- Phase 2: Track - Create or reference a tracking issue in GitHub, Paperclip, or a local file to act as the central source of truth.
- Phase 3: Package Context - Generate and attach structured documents (plan, design, and context) to the tracking issue to ensure agent-readability.
- Phase 4: Route - Decide on the code destination (existing repo, new repo, or local) and create a branch linked to the issue identifier.
- Phase 5: Hand Off - Spawn an executing agent via the sessions_spawn protocol, ensuring full enforcement of linting and PR creation gates.
- Phase 6: Execute - Perform minimal-impact changes following a strict recovery hierarchy if blockers or context resets occur.
- Phase 7: Verify - Validate the work through automated tests, static checks, and runtime smoke tests before closing the tracking loop.
Workflow Execution Setup
To begin using this workflow with your Openclaw Skills, ensure your environment is configured for your preferred tracking system. Use the following structure to initialize a tracking context:
# Navigate to your project and initialize the tracking reference
# The skill is tracker-agnostic (GitHub, Paperclip, or Local)
# Example for local file tracking:
touch .openclaw/tracking/issue-001.md
Refer to the internal references directory within the skill to configure specific API integrations for external platforms like GitHub or Paperclip.
Workflow Execution Data Schema & Taxonomy
The skill organizes metadata and implementation details through a structured document taxonomy attached to the tracking issue:
| Document Key | Role | Requirement |
|---|---|---|
plan |
The master roadmap containing the goal, steps, and verification methods. | Mandatory |
design |
Technical specs for new interfaces, UI/UX flows, or architecture tradeoffs. | Optional |
context |
Background knowledge, rationale for prior decisions, and relevant code patterns. | Optional |
issue_id |
Unique string (e.g., SUP-490) used for branch names and session IDs. | Mandatory |
Workflow Execution Advanced Features
- Autonomous Guardrails: Built-in logic for reproducing and isolating bugs with minimal-touch patches during execution.
- Session Spawn Enforcement: Mandatory use of sessions_spawn for code handoffs to trigger automated linting and PR pipelines.
- Context Resilience: A self-correction loop that forces agents to re-read issue documents after model fallbacks or session compactions.
- Subagent Orchestration: Strategy for spawning specialized subagents with clean ownership boundaries for multi-system work.
SKILL.md
Loading
Related Openclaw Skills
Design Ads
An automated AI design agent that generates high-resolution, technical-style posters and covers using professional design systems.
OpenClaw Compact Improver
A professional optimization tool for the OpenClaw /compact command that implements a 9-part structured summary to ensure high-fidelity context retention.
OpenClaw Memory Freshness
A sophisticated utility for OpenClaw that injects time-based freshness warnings into memory search results to ensure AI agents validate outdated information.
Dream Rem
A sophisticated memory maintenance tool that distills daily logs into thematic topic files while pruning outdated information.
High-Value Task Manager
A strategic task management system that uses headhunting methodologies to prioritize tasks based on quantitative value, energy costs, and conflict detection.
OpenClaw Indian Advocate
A specialized AI legal assistant designed to help Indian advocates navigate procedural law, draft court documents, and manage the transition to post-2024 criminal law reforms.
Featured*
