[refactoring] Extract repo-memory configuration patterns into shared component #23101
Closed as not planned
Description
Skill Overview
The repo-memory tool is used in 26 workflows to persist historical data (metrics, audit results, trends) across workflow runs. Despite serving different use cases, these workflows share a nearly-identical configuration pattern and duplicated prompt guidance around JSON Lines storage, 90-day data retention, and file organization.
Current Usage
The repo-memory: tool appears in:
-
agent-performance-analyzer.md -
audit-workflows.md -
code-scanning-fixer.md -
copilot-agent-analysis.md -
copilot-cli-deep-research.md -
copilot-pr-nlp-analysis.md -
copilot-pr-prompt-analysis.md -
copilot-session-insights.md -
daily-cli-performance.md -
daily-code-metrics.md -
daily-community-attribution.md -
daily-copilot-token-report.md -
daily-news.md -
daily-testify-uber-super-expert.md -
deep-report.md -
delight.md -
developer-docs-consolidator.md -
discussion-task-miner.md -
firewall-escape.md -
glossary-maintainer.md -
metrics-collector.md -
pr-triage-agent.md -
security-compliance.md -
technical-doc-writer.md -
weekly-blog-post-writer.md -
workflow-health-manager.md
Repeated configuration block (nearly identical in 20+ workflows):
repo-memory:
branch-name: memory/<workflow-name>
description: "Historical <metric type> data"
file-glob: ["*.json", "*.jsonl", "*.csv", "*.md"]
max-file-size: 102400 # 100KBRepeated prompt patterns include:
- JSON Lines (
.jsonl) append-only storage format - 90-day data retention / rolling window logic
- ISO 8601 timestamp requirements in data points
- Directory structure guidance (
memory/<workflow>/)
Proposed Shared Component
File: .github/workflows/shared/repo-memory-guidelines.md
Configuration:
---
# No tool config – branch-name is workflow-specific and cannot be shared
# This component provides only prompt guidelines
---Content (prompt section):
## Repo-Memory Data Storage Guidelines
Use `repo-memory` to persist historical data across workflow runs. Follow these patterns for consistency:
### File Format: JSON Lines (Recommended)
Use `.jsonl` (JSON Lines) for append-only time-series data:
\`\`\`python
import json
from datetime import datetime
# Append a data point
data_point = {"timestamp": datetime.utcnow().isoformat() + "Z", "value": 42, ...}
with open('/tmp/gh-aw/repo-memory/default/memory/<workflow>/history.jsonl', 'a') as f:
f.write(json.dumps(data_point) + '\n')
\`\`\`
### Standard File Glob Pattern
Configure `repo-memory` with these standard settings in your workflow's `tools:` section:
\`\`\`yaml
tools:
repo-memory:
branch-name: memory/<your-workflow-name>
description: "Historical <metric type> data"
file-glob: ["memory/<your-workflow-name>/*.json", "memory/<your-workflow-name>/*.jsonl",
"memory/<your-workflow-name>/*.csv", "memory/<your-workflow-name>/*.md"]
max-file-size: 102400 # 100KB
\`\`\`
### 90-Day Data Retention
Always enforce a 90-day retention window to prevent unbounded growth:
\`\`\`python
import pandas as pd
from datetime import datetime, timedelta
cutoff = datetime.utcnow() - timedelta(days=90)
df = pd.read_json('history.jsonl', lines=True)
df = df[pd.to_datetime(df['timestamp']) >= cutoff]
\`\`\`
### Reading Historical Data
\`\`\`python
# Read data from repo-memory
df = pd.read_json('/tmp/gh-aw/repo-memory/default/memory/<workflow>/history.jsonl', lines=True)
df['date'] = pd.to_datetime(df['timestamp']).dt.date
\`\`\`Usage Example:
imports:
- shared/repo-memory-guidelines.mdImpact
- Workflows affected: 26 workflows
- Lines saved: ~20–40 lines of prompt guidance per workflow = ~600 lines total
- Maintenance benefit: When JSON Lines format best practices change or retention policies update, only 1 file needs updating
- Consistency: Currently, some workflows use 30-day retention, others use 90-day, and some have no documented retention policy
Implementation Plan
- Create
.github/workflows/shared/repo-memory-guidelines.mdwith shared usage patterns - Review all 26 workflows to identify which already have JSON Lines patterns and which need guidance
- Add
- shared/repo-memory-guidelines.mdto theimports:list in all 26 workflows - Remove duplicated JSON Lines / retention guidance from workflow bodies
- Standardize all workflows to use 90-day retention windows
- Run
make recompileand test 2-3 representative workflows
Related Analysis
This recommendation comes from the Workflow Skill Extractor analysis run on 2026-03-26.
See the full analysis report in discussions for the complete findings.
Generated by Workflow Skill Extractor · ◷
- expires on Mar 28, 2026, 11:43 AM UTC