Add local cache system for external ticketing/documentation systems

[Jeffallan]

Problem

Workflow commands make repeated API calls to external systems (Jira, Linear, GitHub Issues, Confluence, etc.) during each session. This causes:

• Slow context gathering (19,307 Read operations in usage data, many for external content)
• Session failures when APIs are slow or unavailable
• Redundant fetching of unchanged data

Proposal

Implement a generalized local cache system that works with any supported ticketing/documentation backend:

Cache Structure

.claude/cache/
├── tickets/
│   ├── index.json          # Ticket metadata + relationships
│   └── {ticket-key}.json   # Individual ticket details
├── docs/
│   └── {doc-id}.json       # Cached documentation pages
└── sync-state.json         # Last sync timestamps per source

Behavior

1. On first run: Fetch all relevant data, populate cache
2. On subsequent runs: Read from cache first, sync only changes
3. Manual refresh: /sync-cache command to force full refresh
4. Automatic expiry: Configurable TTL per data type

Backend Abstraction

Cache system should be backend-agnostic:

• Jira adapter: Maps Epic Link relationships
• Linear adapter: Maps project/issue hierarchies
• GitHub adapter: Maps labels/milestones
• Confluence adapter: Maps space/page structure

Benefits

• Eliminates API as workflow bottleneck
• Enables offline/degraded-mode operation
• Faster epic planning (query local JSON vs. API round-trips)

Relates To

• Generalize workflow commands for multiple documentation and ticketing systems #62 Generalize workflow commands for multiple systems
• Context Persistence: Auto-store epic workflow state #50 Context Persistence: Auto-store epic workflow state

[DankCoder20]

Assign the issue to me