- Home
- Skills
- Workflow Automation
- Data Cleaning Guardrails
More screenshots
Works with the AI Agents you already use
Data Cleaning Guardrails
A hard-gated cleaning workflow that never guesses, never silently deletes rows, and always validates before handing data back.
Free
Data Cleaning Guardrails
Example session with this skill installed
Clean this messy CSV I exported from our CRM — some emails are missing and one column looks shifted.
- Read your context and instructions
- Compiled the data cleaning guardrails
The agent profiles the file first (rows, columns, 3-row sample), then outputs a diagnostic report: encoding check, missing counts per column, shifted-column suspicion.
Diagnostic Report
- Missing values: email 12% (14 rows), phone 3%
- Column shift suspected: "contact" column contains mixed zip codes
- Duplicates: 6 exact duplicate rows
Proposed Plan
- Problem: 12% emails missing -> Plan: offer drop / impute / skip -> Impact: 14 rows
- Problem: 6 duplicate rows -> Plan: keep first occurrence -> Impact: 6 rows
- Problem: shifted contact column -> Plan: flag, no auto-fix -> Impact: 1 column
Expected result after cleaning: 860 rows x 18 columns.
The agent then STOPS and waits for explicit confirmation — nothing is modified until you say yes. After cleaning it runs four validation layers (domain / statistical / sampling / shape) and an anomaly checklist (row-count drift > 20%, vanished categories, impossible values, missing rate > 10%) before calling it done.
Connects securely to your tools. The creator never sees your data.
About this skill
Most agents clean data too eagerly: regex-guessing mojibake, auto-dropping rows, and reporting "done" without validation. This skill installs a diagnose -> propose -> confirm workflow with hard gates:
- No cleaning before the user confirms the plan
- Never guess repairs for corrupted fields — flag them instead
- Never auto-delete rows with missing values — present options
- Four-layer validation (domain / statistical / sampling / shape) before any "done"
Pure instruction skill — no dependencies, no scripts to run, works in any agent that follows SKILL.md.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Workflow Automation.
Requires user confirmation at multiple stages. Will not automatically repair undecidable encoding corruption or delete rows without explicit approval.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 3 days ago
- Free with an Agensi account
Nothing to install. It runs in your AI agent.
Nothing to install. The skill is a set of behavioral rules the agent loads from SKILL.md.
Permissions
File Scopes
Instruction-only: the agent reads the user's data files, proposes a cleaning plan, and writes cleaned output only after explicit user confirmation. No scripts, no network access.