[plan] Add external data ingestion pattern for workflows requiring non-GitHub data sources

[github-actions]

Objective

Create a documented pattern and example workflow for ingesting external (non-GitHub) data into agentic workflows, addressing scenarios that require cloud billing metrics, monitoring data, or other out-of-band data sources.

Context

From Agent Persona Exploration discussion #18026:

Add an "external data ingestion" pattern — Scenarios requiring non-GitHub data (cloud billing, monitoring metrics) need a documented ingestion pattern (e.g., writing data to a branch/artifact first, then triggering the agentic workflow).

The worst-scoring scenario (DO-2, 3.8 — Infra Cost Anomaly) failed because it tried to access GitHub billing API for real cloud infrastructure costs (AWS/GCP/Azure), which require external API access. The recommended pattern is a two-phase approach: ingest external data first, then trigger the agentic workflow.

Approach

1. Design a two-phase pattern:
   • Phase 1: A standard GitHub Actions workflow fetches external data (cloud APIs, monitoring endpoints) and stores it as an artifact or in a branch file
   • Phase 2: An agentic workflow triggered by workflow_run consumes the stored data for analysis/action
2. Add documentation in docs/ describing:
   • Why direct external API access from agentic workflows is limited
   • The ingestion pattern with a sequence diagram
   • Network allowlist configuration for the ingestion phase
   • How to pass data between phases (artifacts, branch files, workflow outputs)
3. Create an example workflow pair demonstrating the pattern (e.g., cost anomaly detection or uptime monitoring)
4. Document how to configure network.allowed for the external API endpoints in the ingestion phase

Files to Create/Modify

• docs/ — Add "External Data Ingestion" guide
• .github/workflows/ — Add example workflow pair demonstrating the pattern

Acceptance Criteria

• Documentation explains the two-phase ingestion pattern with clear diagrams/examples
• Example workflow shows how to store external data as artifact and consume it in an agentic workflow
• Network configuration guidance included for the ingestion phase
• Pattern handles common external data sources (REST APIs, webhooks)
• Covers error handling when external data is unavailable

Generated by Plan Command for issue #discussion #18026

• expires on Feb 26, 2026, 3:15 AM UTC

[github-actions]

This issue was automatically closed because it expired on 2026-02-26T03:15:22.910Z.

Closed by Workflow