research-units-pipeline-skills

🥈Silver

Research pipelines as semantic execution units: each skill declares inputs/outputs, acceptance criteria, and guardrails. Evidence-first methodology prevents hollow writing through structured intermediate artifacts.

397170Updated Today

Intermediate30min to implementautomation

Saves ~150 min per use

Quick InstallView Source

git clone https://github.com/WILLOSCAR/research-units-pipeline-skills.git

Works with:

Claude

Overview

About This Skill

How to Use

1. **Prepare the Pipeline Definition:** Extract the pipeline's semantic execution unit (e.g., YAML, JSON, or code comments) and provide it as `[PIPELINE_DEFINITION]`. Include all skills, their inputs/outputs, and guardrails. 2. **Run the Analysis:** Paste the prompt into your AI tool (e.g., Claude, ChatGPT) and replace `[PIPELINE_NAME]` and `[PIPELINE_DEFINITION]` with your specific pipeline. Ensure the pipeline definition is complete to avoid incomplete results. 3. **Review the Report:** The AI will generate a structured report highlighting discrepancies, missing guardrails, and semantic inconsistencies. Focus on the **Validation Status** column to prioritize fixes. 4. **Iterate and Validate:** Address the recommendations, then re-run the analysis to confirm improvements. Use tools like `pytest` or custom scripts to validate guardrails in your CI/CD pipeline. 5. **Document Findings:** Save the report as part of your pipeline's documentation (e.g., in a `PIPELINE_AUDIT.md` file) to track semantic consistency over time. **Tip:** For large pipelines, break the analysis into smaller chunks (e.g., by skill category) to avoid overwhelming the AI. Use the `Validation Status` column to prioritize fixes based on severity (e.g., ❌ for compliance risks, ⚠️ for potential issues).

Setup & Installation

Quick Install

No install command available. Check the GitHub repository for manual installation instructions.

Alternative Install (Git Clone)

git clone https://github.com/WILLOSCAR/research-units-pipeline-skills

Requirements

Claude Code or compatible AI agent
Works with: Claude

Quick Start Guide

Install the Skill

Copy the install command above and run it in your terminal.

Open Your AI Agent

Launch Claude Code, Cursor, or your preferred AI coding agent.

Try It Out

Use the prompt template or examples below to test the skill.

Customize

Adapt the skill to your specific use case and workflow.

Usage Examples

Prompt Template

Analyze the [PIPELINE_NAME] pipeline as a semantic execution unit. Identify all declared skills in the pipeline, including their inputs, outputs, acceptance criteria, and guardrails. For each skill, verify if the declared artifacts are semantically consistent with the pipeline's purpose. Highlight any discrepancies or missing guardrails. Provide a structured report with sections for skills, artifacts, and validation findings. Pipeline definition: [PIPELINE_DEFINITION]

Example Output

### Pipeline Analysis: Customer Onboarding Automation Pipeline

#### **1. Declared Skills in the Pipeline**

| Skill Name | Inputs | Outputs | Acceptance Criteria | Guardrails | Validation Status |
|------------|--------|---------|---------------------|-----------|-------------------|
| **validate_customer_data** | `customer_id`, `email`, `phone` | `is_valid: bool`, `validation_errors: list` | `is_valid` must be `True` if all fields meet format rules | `email` must not be empty, `phone` must be 10 digits | ✅ **Valid** - All artifacts align with criteria. |
| **fetch_credit_score** | `customer_id`, `ssn` | `credit_score: int`, `credit_report: dict` | `credit_score` must be between 300-850 | `ssn` must be masked in logs | ⚠️ **Warning** - `ssn` is not masked in declared artifacts. |
| **generate_contract** | `customer_id`, `credit_score`, `plan_type` | `contract_pdf: file`, `contract_id: str` | `contract_pdf` must be generated within 5 seconds | `plan_type` must be one of: `basic`, `premium`, `enterprise` | ❌ **Invalid** - No performance guardrail for `contract_pdf` generation time. |

#### **2. Semantic Consistency Findings**

- **Discrepancy in `fetch_credit_score`:** The skill declares `ssn` as an input, but the guardrail for masking is not enforced in the output artifacts. This could lead to PII exposure in logs or intermediate storage.
- **Missing Guardrail in `generate_contract`:** The pipeline does not enforce a timeout for contract generation, which could cause delays in the onboarding process.
- **Incomplete Artifacts in `validate_customer_data`:** While the skill declares `validation_errors` as an output, the pipeline does not specify how these errors are propagated to downstream skills (e.g., `notify_support_team`).

#### **3. Recommendations**

1. **Enforce PII Masking:** Update the `fetch_credit_score` skill to ensure `ssn` is never logged or stored in plaintext. Use a secure hashing mechanism for storage.
2. **Add Performance Guardrails:** Introduce a 5-second timeout for `generate_contract` to prevent pipeline stalls. Log failures for SLA tracking.
3. **Define Error Propagation:** Add a `validation_errors` field to the pipeline's global state so downstream skills can react to data quality issues.

#### **4. Pipeline Health Score: 65/100**

The pipeline has **3 critical gaps** that could lead to runtime failures or compliance violations. Addressing these will improve reliability and security.