Przejdź do treści
Biblioteka skilli
Programowanie · Automatyzacja

executing-plans

/executing-plans

Wykonuje zatwierdzony plan w katalogu .workflow: aktualizuje etapy, przydziela zadania, odzyskuje stan po kompaktowaniu i scala wyniki.

Repozytorium
marketplace ↗
Licencja
MIT
Ostatnia zmiana repozytorium
22.10.2026

Ocena i bezpieczeństwo

90/ 100Klasa: dobry
Format25 / 25
Frontmatter z nazwą i opisem, poprawna nazwa, opis do 1024 znaków, nazwa zgodna z folderem.
Odkrywalność25 / 25
Opis mówi, kiedy użyć skilla, ma sensowną długość i jest w 3. osobie.
Treść25 / 35
Co najmniej 150 słów i do 500 linii, nagłówki, listy kroków, przykłady albo kod.
Higiena15 / 15
Bez zaślepek typu TODO, bez ścieżek Windows, z licencją.
Ryzykobrak sygnałów w skanie
Pliki w paczce
2
Skrypty
0
Komentarze HTML
0
Adresy URL w SKILL.md
0
Przestarzałość

Brak sygnałów. Repozytorium zmienione 22.10.2026, bez odwołań do wycofanych modeli i API.

Licencja

MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.

Co robi i kiedy

Wykonuje zatwierdzony plan w katalogu .workflow: aktualizuje etapy, przydziela zadania, odzyskuje stan po kompaktowaniu i scala wyniki.

Oryginalny opis (EN)

Use to execute an approved active .workflow plan with milestone state updates, scoped task ownership, compaction recovery, selective worktree delegation, and one integration path. Do not use for direct fast-path work without a durable plan.

Jak działa

  1. Start Or Resume
  2. Execution Cadence
  3. Semantic Work Routing
  4. Delegation Gate
  5. Evidence Ownership
  6. Completion
Sekcje z nagłówków SKILL.md, w oryginale · 65 linii, 680 słów
Podgląd SKILL.md
---
name: executing-plans
description: Use to execute an approved active .workflow plan with milestone state updates, scoped task ownership, compaction recovery, selective worktree delegation, and one integration path. Do not use for direct fast-path work without a durable plan.
---

# Executing Plans

Execute the active plan as written. Plan artifacts are authoritative; chat is not. Inherit the primary path, assurance profile, task-wide agent budget, and verification budget from `using-workflow`.

## Start Or Resume

1. Call `workflow_status` once when active state is unknown or context was recovered.
2. Read `manifest.json`, `state.json`, `plan.md`, and the current task's referenced context/decisions.
3. Read `ui-contract.md`, chunk state, or older artifacts only when the current task depends on them.
4. Inspect `git status --short` and preserve unrelated user changes.
5. Continue from the first non-complete task; do not replay completed work.

Use `workflow_plan_update`, `workflow_plan_artifact_write`, and `workflow_plan_complete` when available; manual state/artifact writes are fallback only.

Read [references/execution-state.md](references/execution-state.md) only when updating task/chunk state, delegating a write task, or recovering a complex run.

## Execution Cadence

- Update state at meaningful milestones: task start, material block, review handoff, and task completion.
- Do not write workflow state after every tool call or tiny edit.
- Execute tasks within their allowed scope and record any approved scope change before editing outside it.
- Do not perform opportunistic refactors. A new API/schema/protocol/shared abstraction/layer or a higher change class requires an approved plan update and a renewed commitment reflection before implementation continues.
- For production UI, read the UI contract's reuse receipt before markup/styles. Its exact labels are `Target:`, `Shared primitive:`, `Layout precedent:`, `Decision:`, `Structural verification:`. Treat named reuse as architecture acceptance; stop before a local substitute, duplicate primitive, or unapproved layout deviation.
- For a small user correction, inspect and edit only the touched surface plus directly affected tests; do not restart repository discovery.
- For chunks, root owns dependency order and integration. Never create nested chunks.

## Semantic Work Routing

Exact model and reasoning effort live only in canonical custom-agent TOML files.

- `mechanical`: tiny fully specified transformation; `workflow_implementer`.
- `structured`: bounded implementation with locked approach; `workflow_implementer`.
- `standard`: moderate scoped reasoning; `workflow_implementer_standard`.
- `complex`: architecture or broad implementation inside a defined boundary; `workflow_implementer_complex`.
- `critical`: parent retains final decisions; use an independent risk role only within assurance budget.

Do not launch analysis to restate a decision-complete plan.

## Delegation Gate

At each meaningful task boundary, choose local execution or the best bounded agent role deliberately. Prefer one focused agent for nontrivial diagnosis, multi-surface review, broad read-only exploration, or an implementation task that can proceed independently; these are normal standard-path uses, not exceptional escalation. Keep tiny direct fixes and already-decided single-surface work local. Never fan out by file count, checklist length, number of chunks, or number of applicable skills.

All delegated write work uses a worktree. The worker receives goal, chosen approach, allowed scope, exact edits, non-goals, verification, stop-if-unclear behavior, and report format. If a required named role is unavailable, execute locally unless that independent role was explicitly required.

The agent budget is global across planning, execution, and finalization. Reuse a running agent for focused follow-up. A launch used for exploration cannot be replaced by a fresh implementation or review quota.

## Evidence Ownership

The worker runs scoped checks for its assigned change. The parent verifies output exists, performs minimal diff/scope sanity, and runs only integration or acceptance evidence not already proven by an unchanged worker result. Do not rerun the same command solely because ownership returned to the parent.

After relevant edits, rerun the strongest affected check once. Merge delegated work only after the applicable gate passes.

If implementation materially diverges from the reviewed commitment, call `workflow_plan_commitment_propose` with the new candidate and confirm a revised decision before handoff. Hook enforcement is optional and must not be awaited or treated as a dependency.

## Completion

During active user testing, batch small corrections and avoid full verification after each one. When implementation tasks are complete, set phase to `finalizing`.

- `fast` work should not be in this module.
- `standard` and `assurance` continue to `finalizing-plan` with the remaining task-wide budget.
- After final verification/review succeeds, close the run with `workflow_plan_complete`; `set_phase: complete` alone is insufficient.

Stop execution when all planned tasks are complete or a material unresolved decision requires the user. Do not add polish or new scope during handoff.
Pliki w paczce (2)
  • SKILL.md 5 KB
  • references/execution-state.md 2 KB

Wersje

RepozytoriumWynikLicencja
Ta strona · wiolett-industries/marketplace90MIT
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

Migawka bazy z 13.11.2026 · wersja paczki fa027731f51f (skrót SHA-256 całej paczki)

Pobrania w ostatnich 30 dniach: 0 · ranking

Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.

Podobne skille

  • 100execute-plan
    Wykonuje ustrukturyzowane plany zadań z plików w katalogu plans i śledzi status realizacji poszczególnych kroków.
  • 90sr-executing-plans
    Wykonuje spisany plan wdrożenia w osobnej sesji, z punktami kontrolnymi przeglądu.
  • 100implement-plan
    Wykonuje zatwierdzony plan implementacji z katalogu planów, faza po fazie, z weryfikacją.
  • 90evanflow-executing-plans
    Wykonuje zatwierdzony plan wdrożenia zadanie po zadaniu, z weryfikacją na bieżąco, i zatrzymuje się, gdy pojawi się blokada.
  • 100execute
    Realizuje zatwierdzony plan wdrożenia z .ai/plans/ zadanie po zadaniu: uruchamia weryfikację każdego zadania i commituje przez bramkę check-and-commit, a przy luce w planie zatrzymuje się.
  • 100review-plan
    Sprawdza plan implementacji przed jego wykonaniem, zwłaszcza wygenerowany przez writing-plans, zanim ruszy executing-plans.

Uwagi i zgłoszenia

Wiadomość trafi do Bartosza Sokolińskiego. Adres e-mail jest potrzebny tylko do odpowiedzi. Szczegóły w polityce prywatności.