strategy-os: Skill for Claude Code

Decision Enforcement

Apply the mandatory decision rules to the gap ledger and destruction outcomes. Produce state changes, readiness gate values, execution queue entries, and governor escalations.

Every rule must be evaluated. No rule may be skipped. Every block must have explicit rationale.

---

Process

Step 1: Collect Inputs [S]

Gather from prior skill outputs and register state:

1. Gap Ledger (scored and ranked gap records from gap-computing-ledger).
2. Destruction Log (outcomes from gap-running-destruction).
3. Current confidence states for all hypotheses.
4. Current support states for all proposals.
5. Decision deadlines and their statuses.
6. Active blockers.
7. Any governor responses from execution/queue/.

Gate: inputs_collected: bool -- gap ledger and destruction log both available. If either is missing, halt: "Cannot enforce decisions without gap computation and destruction pass."

• Pass: Step 2.
• Fail: halt. Report which input is missing.

Step 2: Apply Priority Rule [S]

Rule: Always work on the gap with the highest (gap score x blast radius).

1. Read the ranked gap list from gap-computing-ledger.
2. Confirm the #1 gap is the current work priority.
3. If the current execution queue contains tasks not aligned with the top gap, flag them.

Gate: priority_confirmed: bool -- top gap identified, priority stated.

• Pass: Step 3.
• Fail: re-rank gaps.

Step 3: Apply Execution Rule [R]

Rule: A task is valid only if it reduces a top-3 open gap and produces T1 or T2 evidence.

For each proposed or active task:

Check	Pass	Fail
Targets a top-3 gap?	Valid	Reject: "does not reduce a top-3 gap"
Produces T1 or T2 evidence?	Valid	Reject: "does not produce qualifying evidence"

List valid tasks and rejected tasks with rejection reason.

Gate: execution_filtered: bool -- every task evaluated. Invalid tasks rejected with reason.

• Pass: Step 4.
• Fail: re-evaluate tasks against current top-3 gaps.