Przejdź do treści
Biblioteka skilli
DevOps i chmura · Automatyzacja

executing-operation-plans

/executing-operation-plans

Wykonuje pisemny plan operacji infrastrukturalnej w osobnej sesji, z punktami przeglądu między krokami.

Repozytorium
srepowers ↗
Licencja
MIT
Ostatnia zmiana repozytorium
09.11.2026

Ocena i bezpieczeństwo

100/ 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.
Treść35 / 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 09.11.2026, bez odwołań do wycofanych modeli i API.

Licencja

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

Co robi i kiedy

Wykonuje pisemny plan operacji infrastrukturalnej w osobnej sesji, z punktami przeglądu między krokami.

Oryginalny opis (EN)

Use when you have a written infrastructure operation plan to execute in a separate session with review checkpoints - for long-running operations requiring human review between steps

Jak działa

  1. Overview
  2. When to Use
  3. The Process
  4. Deviation Handling
  5. Environment Promotion
  6. Stop and Rollback
  7. Integration
Sekcje z nagłówków SKILL.md, w oryginale · 173 linii, 797 słów
Podgląd SKILL.md
---
name: executing-operation-plans
description: Use when you have a written infrastructure operation plan to execute in a separate session with review checkpoints - for long-running operations requiring human review between steps
license: MIT
---

# Executing Infrastructure Operation Plans

## Overview

Load infrastructure operation plan, review critically, execute tasks in batches with verification at each step, report for review between batches.

**Core principle:** Batch execution with checkpoints for safety verification and human review.

**Announce at start:** "I'm using the executing-operation-plans skill to execute this infrastructure operation plan."

## When to Use

| Use This When | vs. subagent-driven-operation |
|---------------|------------------------------|
| Separate session (can pause/resume) | Same session |
| Human-in-the-loop between batches | Subagent executes autonomously |
| Complex multi-environment operations | Single environment |
| High-risk changes need approval checkpoints | Lower risk, faster iteration |

## The Process

### Step 0: Set Up Isolated Workspace

Before loading the plan:

```bash
git branch --show-current
kubectl config get-contexts -o name
```

**REQUIRED:** Use srepowers-core:using-git-worktrees-sre to create an isolated worktree for this operation. Do not execute on the main branch without explicit user consent.

### Step 1: Load and Review Plan

1. Read operation plan file
2. Parse YAML frontmatter: extract `risk_level`, `environment`, `tasks_count`, `status`
3. **Check for resume state:**
   - If `status: "in_progress"` or `status: "completed"` in frontmatter, read `## Execution Status` section
   - Identify completed tasks (marked `[x]`) and pending tasks (marked `[ ]`)
   - Announce: "Resuming operation. Tasks 1-[N-1] completed. Starting from Task [N]."
4. Review critically:
   - Are verification commands clear?
   - Are rollback steps documented?
   - Are environment boundaries respected?
5. If concerns: Raise them before starting
6. If no concerns: Record the execution state and proceed

### Step 2: Pre-Execution Safety Check

Before executing any operations:

```bash
# Verify current environment
git branch --show-current
kubectl config get-contexts -o name

# Check for active incidents
kubectl --context <context> get events --all-namespaces --field-selector type=Warning | head -10
```

**If incidents in progress:** STOP. Report and defer operation.

### Step 3: Execute Batch

**Default: First 3 tasks or first environment**

For each task, follow **test-driven-operation** discipline:
1. RED: Define verification or baseline before execution
2. Verify RED or baseline: Run and watch it fail when target state is absent/broken, or capture baseline when failure is not meaningful
3. GREEN: Execute minimal operation
4. Verify GREEN: Confirm success
5. Check side effects

### Step 4: Batch Verification

After completing batch:

```bash
# Verify changes from this batch
kubectl --context <context> get <resources> -n <namespace>

# Check for errors in affected components
kubectl --context <context> logs -n <namespace> -l <label> --tail=20

# Verify dependent services still healthy
kubectl --context <context> get pods --all-namespaces | grep -v Running
```

**Update execution state in plan file:**
- Mark completed tasks as `[x] completed (commit SHA)`
- Update frontmatter `status: "in_progress"`

### Step 5: Report and Checkpoint

```
Batch X Complete

Tasks: Task N ✅, Task N+1 ✅, Task N+2 ✅

Infrastructure State:
- Environment: <env>
- Components: <status>
- No errors: <verification>

Ready for:
1. Review current state
2. Approve next batch
3. Rollback if issues

Say "continue" to proceed.
Say "rollback" to revert this batch.
```

### Step 6: Continue or Complete

**If more tasks:** Wait for approval, execute next batch

**If all complete:** Final verification, update frontmatter `status: "completed"`, update Requirements Traceability status, use `finishing-operation-branch`

## Deviation Handling

When unexpected situations arise during task execution:

| Rule | Type | Action | Max Retries |
|------|------|--------|-------------|
| **R1 - Minor bug** | Auto-fix | Fix, re-verify, continue | 3 |
| **R2 - Missing info** | Auto-resolve | Read adjacent files, infer, continue | 3 |
| **R3 - Verification drift** | Auto-adapt | Adjust expected output, re-verify | 3 |
| **R4 - Scope/arch change** | **STOP** | Present to human, await approval | N/A |

**After 3 failed R1-R3 retries:** Escalate to R4. Report what was tried and what's needed.

**Scope boundary:** Do not auto-fix pre-existing issues unrelated to the current task.

## Environment Promotion

For multi-environment operations (sit → uat → prod):

| From | To | Verification Required | Approval |
|------|-----|----------------------|----------|
| sit | uat | All checks pass in sit | Standard |
| uat | prod | All checks pass in uat | **Explicit required** — ask user directly: "All UAT checks passed. Confirm promotion to prod?" |

## Stop and Rollback

**STOP immediately when:**
- Verification fails after operation
- Unexpected side effects detected
- Infrastructure becomes unhealthy
- Active incident detected
- Instruction unclear

**Rollback immediately when:**
- Service unavailable
- Error rates increase
- Data integrity issues

**Rollback procedure:**
1. Stop current batch
2. Execute rollback commands from plan
3. Verify rollback success
4. Document what happened
5. Resume only after root cause understood

## Integration

**Required workflow skills:**
- **using-git-worktrees-sre** - Set up isolated workspace
- **writing-operation-plans** - Creates the plan this skill executes
- **test-driven-operation** - Execute each operation with verification
- **finishing-operation-branch** - Complete after all batches
Pliki w paczce (2)
  • README.md 2 KB
  • SKILL.md 6 KB

Wersje

W bazie nie ma innych skilli o nazwie executing-operation-plans.

Migawka bazy z 13.11.2026 · wersja paczki 19e9f71146e9 (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

  • 90executing-plans
    Wykonuje pisemny plan wdrożenia w osobnej sesji, z punktami przeglądu.
  • 90executing-plans
    Wykonuje pisemny plan wdrożenia w osobnej sesji, z punktami kontrolnymi do przeglądu.
  • 85executing-plans
    Wykonuje napisany plan implementacji w osobnej sesji z punktami kontrolnymi przeglądu.
  • 90executing-plans
    Wykonuje spisany plan wdrożenia w osobnej sesji z punktami kontrolnymi do przeglądu.
  • 90sr-executing-plans
    Wykonuje spisany plan wdrożenia w osobnej sesji, z punktami kontrolnymi przeglądu.
  • 90executing-plans
    Wykonuje spisany plan wdrożenia w osobnej sesji z punktami kontrolnymi do przeglądu.

Uwagi i zgłoszenia

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