decision-support

Decision-Support Assembly

Assemble a structured decision briefing from multiple vault sources: prior decisions, active tasks, polaris priorities, and supporting/conflicting notes. This skill is read-only by default — it presents context for the user to decide, never records a decision without explicit request.

Iron Laws

Read-only by default. The briefing pipeline makes no vault writes. A decision note is created only if the user explicitly requests it after reviewing the briefing.

Check result.success after every MCP call. If decision_support or topic_packet fails, stop and surface the error — a partial briefing is worse than no briefing.

Workflow

1. decision_support(topic="<topic>") — aggregate relevant decisions, open tasks, and references from the vault scoped to the topic. This is the foundation of the briefing.

2. Read ztlctl://decision-queue — load recent decisions plus the active work queue. Surfaces what decisions have already been made and what tasks are in flight.

3. Read ztlctl://polaris — load the vault's current mission, ordered priorities, and decision principles.

4. check_alignment(decision="<proposed decision>") — advisory polaris check. Returns relevant_priorities and alignment reasoning. This never blocks — record the result and proceed.

5. topic_packet(topic="<topic>", mode="decision") — decision-mode packet with supporting and conflicting note links, related decisions, and open questions on the topic.

6. Synthesize and present structured briefing:

   • Prior decisions — from steps 1 and 2: what has already been decided here
   • Active context — open tasks and in-flight work from step 2
   • Conflicting signals — notes in step 5 that oppose the proposed direction
   • Polaris alignment — from step 4: which priorities this decision serves
   • Recommended action — your synthesis of all inputs

Optional decision note

Only if the user explicitly asks after reviewing the briefing: create_note(title="Decision: <title>", subtype="decision", body="<briefing summary>")

Do NOT suggest this unless prompted. Do NOT auto-create decision notes.

Distinction from ztl:align

ztl:align is a quick polaris pass/fail check (2 MCP calls). ztl:decision-support is a comprehensive multi-source briefing (5–6 calls). Use align for "is this on-strategy?" and decision-support for "what do I need to know before deciding?"

See references/decision-workflow.md for decision_support output fields, decision-queue resource format, and briefing structure guidance.