seamus-brady

How to orchestrate work across large documents and multi-agent delegations without burning token budgets. Reconnaissance-first, search-then-read, parallel-after-reconnaissance.

Decision criteria for delegating to the writer agent vs synthesising directly in your own response.

When and how to use the document library

How the Coder agent drives an OpenCode session via dispatch_coder — the engineering loop, when to dispatch, and how to verify the result is real (vs. claimed).

Maintain a personal index of interesting things, open questions, and follow-ups across sessions

When and how to render a promoted export to PDF via the export_pdf tool. Reads the writer's draft → export → PDF lifecycle and the operator-facing failure modes when the host is missing pandoc or tectonic.

Index to the territory. Subsystem inventory, the agent graph, cross-subsystem interactions, and invariants. Read this when something unfamiliar shows up.

How to answer an ask_deputy call concisely. Only meaningful in Phase 2+ (long-lived deputies).

How the observer specialist works in Retainer — role, expected inputs, output shape, quality bar, common pitfalls. The observer reads this at the start of every dispatch.

How the Retainer scheduler specialist works — role, expected inputs, output shape, common pitfalls. The scheduler reads this at the start of every dispatch.

When the Retainer cog should delegate to the observer, how to brief it, what to expect back, and how to verify the result. Observer is the knowledge gateway — recent (hot index) and deep (full archive). Read before dispatching to agent_observer.

When the Retainer cog should delegate to the scheduler, how to brief it, what to expect back, and how to verify the result. Read before dispatching to agent_scheduler.

Decision framework for when and how to delegate work to Retainer's specialist agents. Covers the two specialists in this reference cut (observer, scheduler) plus the cog's direct tools.

How dispatching to specialist agents works in Retainer's current shape — the agent_<name> tool shape, what the reply envelope looks like, how errors propagate. Read with `delegation-strategy` (which covers when to delegate).

Recovery patterns in Retainer's current shape for tool errors, agent failures, and cycle abandonment — when to retry, when to narrow, when to escalate. Read after seeing an unexpected error or when planning a multi-step task that might fail partway.

Mechanics of Retainer's current memory layer — what's narrative vs facts vs cases, when each is written vs read, retention semantics, who can see what. Read with `memory-management` (which covers the decision procedure).