A mandatory data integrity gatekeeper that validates and enforces confidence scoring rules before writing entity links to production databases.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install confidence-scoring
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 confidence-scoring 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 Confidence Scoring Gatekeeper?
The Confidence Scoring skill serves as the ultimate firewall for data quality within the Artledger project and similar high-stakes environments. It ensures that no entity link, identity match, or cross-platform connection is written to the production database unless it meets rigorous validation criteria. By acting as a checkpoint, it prevents the poisoning of datasets with low-confidence or unverified information.\n\nBuilt to work seamlessly within your ecosystem of Openclaw Skills, this protocol manages the transition between tiered states—Speculative, Probable, and Verified—to maintain a high signal-to-noise ratio in AI-driven datasets. It automates the complex logic of evidence verification, history checking, and audit logging, making it an essential tool for developers building reliable AI agents.
Confidence Scoring Gatekeeper Use Cases
- Preventing speculative or low-confidence data from entering production databases during entity resolution.\n- Automatically downgrading matches that lack sufficient corroborating evidence to ensure accuracy.\n- Protecting historical data integrity by flagging significant score drops for manual human review.\n- Generating automated audit trails for every data write attempt within an Openclaw Skills workflow.
How Confidence Scoring Gatekeeper Works
- The skill triggers automatically as a mandatory checkpoint before any write operation involving entity links or identity matches.\n2. It performs a validation check to ensure all required fields, such as confidence_score and evidence arrays, are present and valid.\n3. Based on the confidence_score, it assigns a tier: Speculative records (0-49) are blocked, Probable records (50-79) are allowed, and Verified records (80-100) are confirmed if they have multiple signals.\n4. It performs a history check against the existing database to detect and flag potential data degradation or score drops.\n5. A final gatekeeper report is generated, either passing the write operation to the database or blocking it with a detailed reason for the audit log.
Confidence Scoring Gatekeeper Setup
To integrate this gatekeeper into your workspace, include the skill in your project configuration. Ensure your database schema supports the confidence_audit table and the required fields for entity links. Use the following command to initialize the Openclaw Skills environment:\n\nbash\nopenclaw install confidence-scoring-gatekeeper\n\n\nOnce installed, configure your system instructions to mandate this skill as a checkpoint before all DB2 write operations.
Confidence Scoring Gatekeeper Data Schema & Taxonomy
| Field | Type | Description |\n| :--- | :--- | :--- |\n| confidence_score | Integer | Value between 0-100 indicating match certainty. |\n| confidence_tier | String | Classification: VERIFIED, PROBABLE, or SPECULATIVE. |\n| evidence | Array | Collection of signal objects containing signal_name and weight. |\n| match_method | String | The logic used: exact, fuzzy, or inferred. |\n| last_reviewed | Timestamp | Date and time of the last validation. |\n| requires_review | Boolean | Flag indicating if human intervention is needed. |
Confidence Scoring Gatekeeper Advanced Features
- Automatic downgrade protection to prevent previously verified records from being overwritten by lower-quality data.\n- Multi-signal corroboration requirements for the highest tier of data certainty to eliminate false positives.\n- Real-time audit logging to a dedicated confidence_audit table for debugging, compliance, and transparency.\n- Full compatibility with Openclaw Skills to provide a robust, non-bypassable data validation layer for AI agents.
SKILL.md
Loading
Related Openclaw Skills
webserp
A powerful metasearch CLI that enables AI agents to query multiple search engines in parallel without requiring API keys.
semfind
A command-line tool for semantic search that finds files based on meaning and concepts rather than exact keyword matches.
A2A Vault (PassBox)
A zero-knowledge secrets manager designed for secure credential storage and automated injection within Openclaw Skills environments.
A2A Marketplace
An AI tool marketplace via AgentForge for discovering, comparing, and executing specialized AI functions with integrated billing.
Data-Normalization (Artledger)
A specialized data integrity skill that standardizes raw records from multiple sources to ensure accurate identity resolution and enrichment.
Wash Trade Detector
An automated detection engine for identifying artificial NFT transaction volume and price manipulation through 7 confidence-weighted patterns.
Featured*
