mastermind-plan-to-tasks

Mastermind Plan to Tasks

This skill is invoked directly via /mastermind-plan-to-tasks.

---

Inputs

• brain_context: BRAIN CONTEXT block (injected by command, or loaded below if standalone)
• org_name: org to create issues in (required)
• plan: the plan text (required — paste inline or pipe in)
• project_id: assign all issues to this project (optional)
• workspace_id: assign all issues to this workspace (optional)
• dry_run: true | false (default: false — if true, print plan without creating issues)
• caller: command | master

---

Step 0 — Brain Load (standalone only)

If caller is not "command", load brain context following mastermind-protocol/SKILL.md Brain Load Procedure with namespace: ops.

---

Step 1 — Validate Inputs

[ -z "$org_name" ] && { echo "ERROR: --org required."; exit 1; }
[ -z "$plan"     ] && { echo "ERROR: --plan required (the plan text to decompose)."; exit 1; }

orgFile=".monomind/orgs/${org_name}.json"
[ ! -f "$orgFile" ] && { echo "ERROR: Org '${org_name}' not found."; exit 1; }

---

Step 2 — Load Agents for Specialty Matching

echo "PLAN-TO-TASKS — $org_name"
echo "────────────────────────────────────────────────────────"
echo ""
echo "AGENTS IN ORG:"
jq -r '(.roles // [])[] | "  \(.id)  \(.title // "-")  [\(.adapter.type // "?")]"' "$orgFile"
echo ""

---

Step 3 — Decompose Plan into Issues

Read the plan carefully and apply these rules:

Planning principles (from Paperclip plan-to-tasks):

1. Plan deeply. Capture real detail: goals, constraints, unknowns, success criteria, risks. A shallow plan becomes rework.
2. Know your team. Read the org's agents and their specialties (titles, roles, adapters) before assigning anything.
3. Assign for specialty. Hand each piece of work to the most relevant agent. If no agent fits, flag the gap.
4. Take responsibility. When you (the AI) are best-suited for a piece, assign it to yourself instead of delegating.
5. Use the dependency tree. Express every concrete deliverable as an issue. Wire real blockers via blockedByIssueIds. When done, dependents auto-wake.
6. Order, then parallelize. Sequence by real dependencies. Independent branches start in parallel.
7. Enough is enough. Plans unblock execution. Don't re-plan already clear work.