A specialized AI agent skill designed to trace, audit, and validate end-to-end data flow and state persistence across application layers.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install data-flow-review
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 data-flow-review 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 Data Flow Review?
The Data Flow Review skill is a powerhouse component of Openclaw Skills, engineered to analyze codebases by mapping how data moves dynamically across system boundaries. Instead of judging files in isolation, this skill empowers developers and AI agents to systematically trace data from its initial entry point down to its final persistence layer.
By deploying this asset within Openclaw Skills, you ensure strict correctness during audits, patch updates, and end-to-end flow reconstructions. It prevents common pitfalls like state drift, stale reads, unsafely mixed semantic fields, and unhandled failure states in asynchronous operations.
Data Flow Review Use Cases
- Performing thorough code reviews focused on state consistency and semantic correctness across multi-layered architectures.
- Executing end-to-end data flow reconstruction across distributed files, databases, and microservices.
- Verifying patch safety to guarantee minor changes do not accidentally shift persistence timing or disrupt branch behavior.
- Conducting API field provenance checks across the UI, application store, backend handler, message queue, and downstream consumers.
- Auditing transaction boundaries to ensure retries, cancels, refreshes, or failures do not leave dirty state behind.
How Data Flow Review Works
- Locate Entry Point: Identify the initial trigger of the data, such as a UI page action, route handler, job consumer, event payload, or API endpoint.
- Map the Chain: Build a step-by-step path trace detailing transition steps: entry -> transform -> store/cache/db write -> request/event -> downstream consumer -> next state.
- Isolate Branches: Explicitly mark and map key branches like success vs. failure, retry vs. first attempt, and sync vs. async operations.
- Audit State Boundaries: Inspect every point where data crosses a boundary (e.g., local state to global store) to check for caching issues or stale data.
- Compile Findings First: Generate a prioritized review document highlighting issues by severity, referencing exact lines of code, trigger scenarios, and risks.
Data Flow Review Setup
Integrating the Data Flow Review skill into your workflow is highly straightforward. Add the markdown skill to your configuration folder to enable your AI coding assistant to utilize it.
# Navigate to your agent workspace skills directory
cd ~/.openclaw/skills/
# Create the data flow review skill configuration file
cat << 'EOF' > data-flow-review.md
# Paste the SKILL.md rules here to configure the agent
EOF
# Verify the skill is loaded by listing active Openclaw Skills
openclaw skills list
Data Flow Review Data Schema & Taxonomy
The skill maintains semantic mapping and outputs diagnostic reports structured around canonical entities and flow steps.
Entity Classification Table
| Category | Field Types Tracked | Key Metrics to Record |
|---|---|---|
| Identifiers | IDs, Keys, Slugs, Trace IDs, Correlation IDs | Source, Transformation, Persistence, Consumers, Required Meaning |
| Actors | User IDs, Tenant IDs, Session IDs, Operator IDs | Origin, Drift risk across boundary, Expiry |
| State Control | Status, Phase, Mode, Flags, Versions | Mutex properties, Async resume capability |
| Business Data | Amount, Currency, Plan IDs, Order IDs, Record IDs | Source of truth, Read-write parity |
Findings Output Schema
- Severity: Critical, Major, or Minor.
- Reference: Exact filename and line numbers.
- Trigger Scenario: Path, branch, or interaction model.
- Risk Profile: Analysis of dirty state, stale reads, or contract breakage.
Data Flow Review Advanced Features
- Cross-Boundary Drift Protection: Identifies micro-drift of variable definitions as data moves between memory, caching, database rows, and network payloads.
- Failure and Cancel Path Audits: Ensures partial-success operations clean up transient database changes or local store states during cancellations.
- Semantic Equivalence Verification: Flag-checks expressions like "a || b" to verify that fallback operations do not mask underlying conceptual mismatches.
- Multi-Agent State Synchronization: Coordinates with other Openclaw Skills to maintain unified semantic standards during major multi-file refactorings.
SKILL.md
Loading
Related Openclaw Skills
Models: Attract Women Through Honesty Skill
An AI agent skill that translates Mark Manson's bestselling framework into an interactive, vulnerability-based dating and self-improvement assistant.
Midnight in Chernobyl Study & Research Toolkit
An executable knowledge toolkit based on Adam Higginbotham's "Midnight in Chernobyl" that enables AI agents to analyze the design flaws, human cost, and cover-up of the 1986 disaster.
Intent Engine
Intent Engine is an AI-driven intent recognition and intelligent routing system with a built-in Web management dashboard, providing accurate prompt classification for multi-agent workflows.
Medical Medium Cleanse to Heal
An AI-powered wellness protocol agent designed to guide users through Anthony William's 'Medical Medium Cleanse to Heal' framework, providing precise detox and dietary strategies.
McKinsey 100-Year Knowledge Base
An AI-driven strategic consulting knowledge base equipping agents with McKinsey's core methodologies, thinking models, and comprehensive industry insights.
Shuangwen Rebirth Simulator
An interactive, AI-driven role-playing simulator that lets users experience the thrill of reincarnation and face-slapping victories in classic web novel style.
Featured*
