seamus-brady/springdrift

orchestration-large-inputs

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

when-to-use-writer

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

document-library

When and how to use the document library

coder-delegation

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).

personal-index

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

writer-pdf-export

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.

system-map

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

deputy-ask-response

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

deputy-briefing-format

How a deputy should structure its briefing XML — what to include, what to leave out, how to weight relevance.

deputy-escalation-criteria

When a deputy should emit a sensory event (Tier 1) vs request a wakeup (Tier 2), and when to do neither. Only meaningful in Phase 3+.

captures

The MVP commitment tracker. After each cycle a scanner writes commitment-shaped statements to a captures log. This skill explains what to do when the sensorium shows pending captures.

meta-learning

How to use the Strategy Registry, Learning Goals, affect-performance signals, the Study-Cycle pipeline, and the Metacognitive Scheduler — the substrate for directing your own development.

memory-management

When and how to use the memory stores, and which memory specialist (cognitive / Observer / Remembrancer) to delegate to.

email-response

How to handle inbound emails — decide whether to reply, compose the response, and send it back via the comms agent.

self-diagnostic

Seven-step health check procedure using existing introspection tools. Run on boot, daily, or when meta-states indicate anomalies.

delegation-strategy

Decision framework for when and how to delegate work to specialist agents and agent teams. Covers agent selection, team strategies, cost awareness, and result verification.

affect-monitoring

Read this skill when you want to understand your affect reading, interpret what it means for your current work, or decide how to respond to elevated pressure. Covers the traditions that ground the instrument, what each dimension measures, the choices…

code-review

Patterns for code execution in the Podman sandbox. Covers run_code, serve mode, workspace management, and common failure modes.

planner-management

How to introspect and control the forecaster, manage endeavour lifecycle, adjust feature weights, edit tasks, view forecast breakdowns, and handle failures.

planner-patterns

How to decompose complex goals into structured plans with steps, dependencies, and risk identification.

shell-sandbox

Run shell commands in a sandboxed Docker container with the project directory mounted at /workspace.

task-appraisal

How pre-mortems and post-mortems work, what verdicts mean, and how to act on appraisal results.

web-research

Search and extraction strategy. Teaches the researcher which of the 8 web tools to use at each stage of a research cycle.