// Controlled plan execution with human review checkpoints - loads plan, executes in batches, pauses for feedback. Supports one-go (autonomous) or batch modes.
| name | ring:executing-plans |
| description | Controlled plan execution with human review checkpoints - loads plan, executes in batches, pauses for feedback. Supports one-go (autonomous) or batch modes. |
| license | MIT |
| compatibility | opencode |
| metadata | {"trigger":"Have a plan file ready to execute, want human review between batches","skip_when":"Same session with independent tasks, no plan exists, plan needs revision","sequence_after":"ring:writing-plans, ring:pre-dev-task-breakdown"} |
Load plan, review critically, choose execution mode, execute tasks with code review.
Core principle: User chooses between autonomous execution or batch execution with human review checkpoints.
Two execution modes:
Announce at start: "I'm using the ring:executing-plans skill to implement this plan."
⚠️ THIS STEP IS NON-NEGOTIABLE. You MUST use AskUserQuestion before executing ANY tasks.
Ask: "How would you like to execute this plan?" Options: (1) One-go (autonomous) - all batches with code review, no human review until completion (2) Batch (with review) - pause for human review after each batch
Based on response: One-go → Steps 3-4 loop until done | Batch → Steps 3-5 loop
This is a structural checkpoint, not optional UX polish.
User saying "don't wait", "don't ask questions", or "just execute" does NOT skip this step because:
Common Rationalizations That Mean You're About to Violate This Rule:
| Rationalization | Reality |
|---|---|
| "User intent is crystal clear" | Intent is not the same as explicit selection. Ask anyway. |
| "This is contextual guidance, not absolute law" | Wrong. It says MANDATORY. That means mandatory. |
| "Asking would violate their 'don't ask' instruction" | AskUserQuestion is a 3-second structural gate, not a conversation. |
| "Skills are tools, not bureaucratic checklists" | This skill IS the checklist. Follow it. |
| "Interpreting spirit over letter" | The spirit IS the letter. Use AskUserQuestion. |
| "User already chose by saying 'just execute'" | Verbal shorthand ≠ structured mode selection. Ask. |
If you catch yourself thinking any of these → STOP → Use AskUserQuestion anyway.
If plan has tasks with target: and working_directory: fields:
Track current module:
current_module = None
current_directory = "."
Before each task, check for context switch:
IF task.target != current_module AND current_module != None:
# Prompt user for confirmation
AskUserQuestion:
question: "Switching to {task.target} module at {task.working_directory}. Continue?"
header: "Context"
options:
- label: "Continue"
description: "Switch to {task.target} and execute task"
- label: "Skip task"
description: "Skip this task and continue with next"
- label: "Stop"
description: "Stop execution for manual review"
IF answer == "Continue":
current_module = task.target
current_directory = task.working_directory
ELIF answer == "Skip":
Mark task as skipped → proceed to next
ELSE:
Stop execution → report progress
Load module-specific PROJECT_RULES.md:
IF {task.working_directory}/PROJECT_RULES.md exists:
Instruct agent to read module-specific rules
Module rules override root rules
Pass working directory to agent:
Task(
subagent_type=task.agent,
model="opus",
prompt="Working directory: {task.working_directory}
Before executing, cd to the working directory:
cd {task.working_directory}
If PROJECT_RULES.md exists in this directory, read and follow it.
{task.prompt}"
)
Optimization: To minimize context switches, batch tasks by module when possible:
Default: First 3 tasks
Agent Selection: Backend Go → ring:backend-engineer-golang | Backend TS → ring:backend-engineer-typescript | Frontend → ring:frontend-bff-engineer-typescript | Infra → ring:devops-engineer | Testing → ring:qa-analyst | Reliability → ring:sre
For each task: Check context switch (Step 2.5) → Mark in_progress → Dispatch to agent with working_directory → Follow plan steps exactly → Run verifications → Mark completed
After each batch, REQUIRED: Use ring:requesting-code-review (all 6 reviewers in parallel)
Handle by severity:
TODO(review): [Issue] ([reviewer], [date], Low)FIXME(nitpick): [Issue] ([reviewer], [date], Cosmetic)Proceed when: Zero Critical/High/Medium remain + all Low/Cosmetic have comments
One-go mode: Log internally → proceed to next batch → report only at completion Batch mode: Show implementation + verification + review results → "Ready for feedback." → wait → apply changes → proceed
Use finishing-a-development-branch to verify tests, present options, execute choice.
STOP immediately: Blocker mid-batch | Critical gaps | Unclear instruction | Verification fails repeatedly. Ask rather than guess.
AskUserQuestion for execution mode - NO exceptions* agents over general-purpose when availableOrchestrates the 6-gate development workflow for implementing tasks. Manages state, dispatches specialist agents, and enforces gate requirements.
Captures metrics and learnings from completed development cycles. Enables continuous improvement by analyzing patterns across cycles.
Gate 4 of development cycle - dispatches 6 specialized reviewers (code, business-logic, security, test, nil-safety, consequences) in parallel for comprehensive code review feedback.
Guide to the dev-team skills - specialist developer agents and 6-gate development cycle. Provides backend engineers (Go/TypeScript), frontend engineers, DevOps, SRE, and QA agents.
Skill discovery workflow and agent dispatch patterns for every conversation.
Generate beautiful, self-contained HTML pages that visually explain systems, code changes, plans, and data. Use when the user asks for a diagram, architecture overview, diff review, plan review, project recap, comparison table, or any visual explanation of technical concepts. Also use proactively when you are about to render a complex ASCII table (4+ rows or 3+ columns) — present it as a styled HTML page instead.