Chapter 1: Framework options

Purpose: establish that durable run ledgers, checkpoints, issue-to-PR handoffs, and agent supervision are common patterns; we should avoid inventing more than necessary.

Summary

The initial options report compared a thin custom OpenClaw bridge, Codex-first execution, GitHub-native coding agents, Paperclip-style orchestration, and broader agent frameworks.

OptionConclusion
Thin OpenClaw bridgeUseful for immediate Telegram/GitHub glue, but insufficient as a long-running agent platform.
Codex-firstStrong executor for PM analysis and development, but needs durable orchestration around it.
GitHub-native coding agentsGood for issue-to-PR coding; less complete for cross-repo PM governance.
Paperclip-style orchestrationBest conceptual match for agent teams, budgets, tickets, heartbeats, and approvals.

Source report: v1 options report