Executing Plans
Load beads epic, review critically, execute tasks in dependency-aware batches, report for review between batches.
Announce at start: "I'm using the executing-plans skill to implement this plan."
REQUIRED BACKGROUND: Read
SKILL.md first (bd CLI usage, permission avoidance, dependency management).
Prerequisites: Beads epic exists (via plan2beads), dependencies set (
shows expected blockers).
The Process
Step 1: Load and Review Epic
, check
and
. Review critically -- raise concerns before starting.
Step 2: Execute Batch
Batch = all currently ready issues (no blockers). For each:
bd update <id> --status=in_progress
then
- Follow each step exactly; run verifications as specified
- REQUIRED: -- any file >50 lines changed -> apply the appropriate rule-of-five variant:
Skill(superpowers-bd:rule-of-five-code)
for code, Skill(superpowers-bd:rule-of-five-tests)
for test files. Complete all 5 passes before proceeding
- Commit the work
- -- unblocks dependent issues
File conflicts: serialize issues touching same file. See
references/batch-execution-detail.md
.
Step 3: Report (Task-Tracked)
Create checkpoint task after each batch. Show: implemented work, verification output, rule-of-five variant passes, newly unblocked issues (
). Say: "Ready for feedback." See
references/report-and-feedback.md
.
Step 4: Continue (Task-Tracked)
Create blocked feedback + execution tasks to enforce checkpoint. Do NOT proceed until human feedback received. See
references/report-and-feedback.md
.
Step 5: Complete Development
All epic issues closed -> verify with
-> hand off to
superpowers-bd:finishing-a-development-branch
. See
references/completion-protocol.md
.
Dependency-Aware Batching
Batch 1: All tasks with no blockers (bd ready)
Close completed -> check bd ready again
Batch 2: Newly unblocked tasks
Close completed -> check bd ready again
See
references/batching-example.md
for full walkthrough.
Epic Scoping
shows ALL epics. Filter to current epic:
for child IDs, only work on issues that are both ready AND children of your epic.
When to Stop
STOP when: blocker mid-batch, no ready issues but open issues remain, unclear instructions, repeated verification failures. Check
for deadlocks. See
references/epic-scoping-and-deadlock.md
.
Ask for clarification rather than guessing.
Remember
- Check before each batch, close issues promptly to unblock dependents
- Follow issue steps exactly, don't skip verifications
- Reference skills when issue says to
- Between batches: report and wait for feedback
- Stop when blocked, don't guess
- REQUIRED: Apply appropriate rule-of-five variant before committing any file >50 lines changed ( for code, for tests)
Integration
- plan2beads -- creates the epic (must run first)
- superpowers-bd:finishing-a-development-branch -- after all issues
- superpowers-bd:subagent-driven-development -- alternative: same session, parallel execution
Reference Files
references/batch-execution-detail.md
: Full Step 2 procedure with file conflict handling and rule-of-five-code trigger
references/report-and-feedback.md
: Steps 3+4 TaskCreate blocks and feedback enforcement flow
references/batching-example.md
: Dependency-aware batching walkthrough with bd ready/blocked at each stage
references/epic-scoping-and-deadlock.md
: Epic scoping, deadlock detection, bd failures, when to revisit
references/completion-protocol.md
: Step 5 completion handoff to finishing-a-development-branch
<!-- compressed: 2026-02-11, original: 652 words, compressed: 434 words -->