Band of Agents — Post-Merge Delivery Pipeline

Code is merged. Now what? Documentation goes stale, testing gets deprioritized, releases get delayed. Existing AI tools handle planning, coding, and review — but nobody handles what comes after the merge. Band of Agents fixes this with four specialist agents that collaborate through Band's platform to automate the post-merge pipeline: 1. Pipeline Orchestrator — receives tasks, discovers available agents via band_lookup_peers, invites them via band_add_participant, shares coordination reasoning via band_send_event 2. Technical Writer — generates API docs, README updates, and docstrings from code changes, then @mentions QA Strategist when docs are ready 3. QA Strategist — analyzes risk areas, generates test plans with edge cases, stores risk assessments via band_store_memory, then @mentions Release Coordinator with a verdict 4. Release Coordinator — reads QA's stored memories via band_list_memories, generates changelogs with semantic versioning proposals, and confirms with TW and QA before shipping This is not a thin wrapper or sequential notification chain. Agents discover each other dynamically, create side rooms for focused discussions, share structured thought events, and persist cross-agent context through Band's memory APIs. Each handoff is visible and auditable in Band chat. Tech stack: LangGraph agents via band-sdk[langgraph], Qwen 2.5 14B (Featherless AI) as primary LLM, Gemini 2.0 Flash as fallback, Python 3.13 with uv for isolated dependency management. Why it matters: Post-merge work is a real enterprise pain point. Teams manually handle documentation, testing, and releases after code is merged — and it is exactly the kind of workflow that benefits from genuine multi-agent collaboration, not just linear automation.