- Home
- Skills
- Research & Analysis
- Evidence Matrix Validator
Evidence Matrix Validator
Validate CSV/TSV evidence matrices for structural integrity, provenance anchors, and identifier consistency.
Free
Works with the AI Agents you already use
Evidence Matrix Validator
Example session with this skill installed
Run scripts/validate_evidence_matrix.py on evidence_matrix.csv and generate a JSON report named qa-audit.json. Summary count only please.
- Read your context and instructions
- Compiled the evidence matrix validator
- Generated the data export
I have executed the evidence matrix validation script against evidence_matrix.csv. The validation process checked for identifier integrity, provenance anchors, and field status consistency.
Validation Summary
| Metric | Value |
|---|---|
| Total Rows Processed | 142 |
| Validation Status | FAILED (Exit Code 1) |
| Error Count | 3 |
| Warning Count | 12 |
| Report Path | qa-audit.json |
Field Status Distribution
- reported: 98
- derived: 22
- not_reported: 10
- conflicting: 5
- unclear: 7
Critical Errors Detected
- Duplicate Identifiers:
record_idERR-042 and ERR-089 appear multiple times. - Missing Provenance: 1 row marked
reportedis missing asource_anchor. - Incomplete Derivation: 2 rows marked
derivedlack the requiredderivation_formulacolumn values.
Next steps
- Open
qa-audit.jsonto identify the specific row indices for the duplicaterecord_iderrors. - Manually verify the source anchors for the flagged rows in
evidence_matrix.csv. - Provide the missing
derivation_inputsfor the rows flagged with statusderived.
evidence_matrix.csv
CSV · data export
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
Evidence matrices often fail during review due to missing provenance, broken identifiers, or inconsistent status flags. Manual inspection of hundreds of rows for structural integrity is error-prone and slow.
This skill validates CSV or TSV evidence matrices against a strict structural contract. It ensures every data point is anchored to a source, validates identifier uniqueness, and checks derived-value logic. It provides immediate feedback on record traceability without altering your source files.
What it does
- Identifier validation checks for duplicate record IDs and ensures required fields like study_family_id and report_id are populated.
- Provenance anchoring verifies that every reported or derived row contains a source_anchor for traceability.
- Status enforcement restricts field_status entries to a specific set of allowed values like reported, derived, or conflicting.
- Derivation checking flags derived rows that lack required derivation formulas or input metadata.
- Structural reporting generates summary counts and optional JSON artifacts for automated QA workflows.
How it works
- Run the validator on a target CSV or TSV file using the provided validation script.
- Specify delimiters and output paths for machine-readable JSON reports if needed.
- Review the terminal summary for row counts, error totals, and specific warning flags.
- Export the QA report for human adjudication or systematic cleanup of the source matrix.
Frameworks & tools
This skill uses a Python-based validation CLI and supports CSV and TSV formats. It integrates with existing data pipelines via standard exit codes and JSON reporting.
Why this beats prompting it yourself
Generic LLM prompts often hallucinate structural fixes or miss subtle identifier duplicates in large datasets. This skill enforces a rigid schema and logic gate, ensuring 100% technical compliance with evidence synthesis standards that a standard chat interface cannot reliably verify.
Use cases
- Validate evidence tables before handing them off for clinical adjudication.
- Audit systematic review matrices for missing provenance anchors.
- Check batch-extracted data for identifier collisions and status consistency.
- Generate QA artifacts for reproducibility documentation in research papers.
Known limitations
The validator checks structure and traceability only. It does not extract data, judge study quality, perform meta-analysis, or verify the clinical accuracy of the data points.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Research & Analysis.
Validates structure and traceability only. Does not judge study quality, verify clinical accuracy, or perform data extraction. Requires specific identifier columns to be present.
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
1 use
Times used to date
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 27 days ago
- Free with an Agensi account