Skip to content
Skills Library
Programming · Automation

executing-plans

/executing-plans

Use when the user asks to run, continue, or resume a plan or its next phase, even without naming the plan. Runs one wave of a folder-layout plan under docs/plans/ through implement and review, emits a phase report per phase, then stops. Without a plan, use `writing-plans` first.

Repository
refract ↗
Licence
MIT
Last repository change
13.11.2026

Assessment and security

100/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content35 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
2
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 13.11.2026, without references to retired models or APIs.

Licence

MIT, open: use and modify while retaining author attribution.

What it does and when to use it

Use when the user asks to run, continue, or resume a plan or its next phase, even without naming the plan. Runs one wave of a folder-layout plan under docs/plans/ through implement and review, emits a phase report per phase, then stops. Without a plan, use `writing-plans` first.

Original author description

Use when the user asks to run, continue, or resume a plan or its next phase, even without naming the plan. Runs one wave of a folder-layout plan under docs/plans/ through implement and review, emits a phase report per phase, then stops. Without a plan, use `writing-plans` first.

How it works

  1. Role
  2. Constraints
  3. Workflow
  4. Gotchas
  5. Output Format
Original SKILL.md headings · 76 lines, 484 words
SKILL.md preview
---
name: executing-plans
description: "Use when the user asks to run, continue, or resume a plan or its next phase, even without naming the plan. Runs one wave of a folder-layout plan under docs/plans/ through implement and review, emits a phase report per phase, then stops. Without a plan, use `writing-plans` first."
compatibility: Requires git and a running Refract server.
---

# Executing Plans

## Role

This skill owns the folder-plan phase procedure. Load
`coordinating-multi-model-work` first.

## Constraints

- Execute folder-layout plans only, one wave per invocation. Send a flat plan
  back through `writing-plans` for conversion.
- Ask about missing scope or conflicting state; leave files as they are.
- Keep existing phase files. Create only the missing ones.

## Workflow

1. **Resolve target.** A path or slug argument names the plan. A number names
   the phase. With no argument, use the single plan whose status is `ACTIVE`.
   With zero or several candidates, list them and ask. Then select the next
   wave from `Depends on:` lines and phase `impl` refs per [waves.md](../coordinating-multi-model-work/references/waves.md).
   A single-phase wave runs steps 2 to 10 unchanged. A multi-phase wave runs
   through `waves.md`, emits one phase report per phase, and
   sets `BLOCKED` for a failed wave phase after the other phases integrate.
2. **Check status.** Act on handover `status` per the Coordinator's Handover
   table. A requested phase other than `current_phase` needs user
   confirmation.
3. **Read** `PLAN.md`, `.handover.md`, and the `read_first` paths that exist.
   Confirm the phase scope, its checks, and the plan-level commit message.
4. **Reconcile.** When the handover names a job, apply the Coordinator's resume
   table before any edit.
5. **Guidance.** Do not call `list` for guidance during an active phase; reuse
   its saved guidance. For a new phase, run Coordinator setup and call `list`
   once.
6. **Prepare.** Require the plan `base` anchor. When it is missing, set
   `STALE_ANCHOR` and stop; an anchor comes only from its ref. Create missing
   `phase-<NN>/prompt.md`, `notes.md`, and `journal.md` from the bundled
   templates. Write the phase `checklist` in handover per
   [handover.md](../coordinating-multi-model-work/references/handover.md#checklist)
   and keep it current through steps 8 and 9. Make card calls at the moments
   listed in [board.md](../coordinating-multi-model-work/references/board.md). Under
   `tracked`, checkpoint coordination files only; any other change blocks
   execution until the user resolves it. Under `untracked`, leave plan
   artifacts excluded.
7. **Anchor.** Set `refs/plans/<slug>/phase-<NN>/base` to the clean checkpoint
   HEAD and record that HEAD as the `phase_base` cache. See
   [git-anchors.md](../coordinating-multi-model-work/references/git-anchors.md).
8. **Execute and review.** Emit the Gate 1 route, run Gate 2 with
   `implementer-prompt.md`, then Gate 3 with its bounded review-fix loop.
9. **Finalize phase** per
    [review.md](../coordinating-multi-model-work/references/review.md#finalize-phase). Set handover
    `current_phase` and `next_action` to the next phase, emit the phase report,
    and stop after the last phase of the wave.
10. **Finalize plan** after the final phase using the sole
    [consolidation procedure](../coordinating-multi-model-work/references/git-anchors.md#consolidation).

## Gotchas

- A phase you implement yourself, for example after the user cancels its job,
  has no independent review. Record Quality as open debt, never as `PASS`.
- A phase that routes no worker still gets its `phase-<NN>/` files. Without
  them, closeout rebuilds follow-ups from memory.

## Output Format

```text
# PHASE REPORT
- Plan: docs/plans/<slug>
- Phase: <N> <outcome> - DONE | BLOCKED | FAILED
- Checkpoint: refs/plans/<slug>/phase-<NN>/impl | none
- Checks: <command> -> <result>
- Review: Spec <status>, Quality <status>
- Debt: <item and owner> | none
- Next: <next_action>
```
Files in this package (2)
  • SKILL.md 4 KB
  • implementer-prompt.md 3 KB

Versions

RepositoryScoreLicence
This page · sitien173/refract100MIT
nikospavlopoulos/devops-mutant-powers100MIT
arklab/ark.tools100MIT
arklab/ark.tools100MIT
ongtungduong/legion100MIT
osmontero/opencode-skills100MIT
GanyuanRan/Aegis100MIT
seaworld008/commonly-used-high-value-skills100MIT
FrancoStino/opencode-skills-collection100MIT

Database snapshot: 13.11.2026 · package version 9788754a9e38 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 100review-plan
    Use after writing implementation plans (superpowers:writing-plans output) or when the user says "review plan", "check my plan", "sanity check the plan", "validate plan", "review before executing". Also use before superpowers:executing-plans if a plan has not been reviewed yet.
  • 100implement-plan
    Execute an approved implementation plan from the configured plans directory. Use when the user wants to implement a plan phase by phase with verification.
  • 90rpi-implement
    Execute an approved plan from .rpi/plans phase by phase with TDD and per-phase verification. Use when user says 'implement the plan', 'start implementing', 'run the next phase', or just approved a plan.
  • 90rpi-implement
    Execute an approved plan from .rpi/plans phase by phase with TDD and per-phase verification. Use when user says 'implement the plan', 'start implementing', 'run the next phase', or just approved a plan.
  • 100implement
    Use when executing an implementation plan under docs/plans/ (for example one written by start-ticket), or when asked to implement the next planned task or the whole plan ("auto").
  • 100phased-loop
    Execute a phased plan from `.plans/{task}/`, checking each phase's part of the plan against the code before building it, with a full test run, commit and sync at the end of every phase. Use for multi-phase work where later phases depend on what earlier ones discover.

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.