
Task Loop
- 25 installs
- 10 repo stars
- Updated July 24, 2026
- duyet/claude-plugins
Runs an iterative task loop where the agent repeatedly works, checks progress, and continues until a task is complete.
About
A Claude Code skill that drives an iterative task loop - the agent works, checks its progress, and keeps going until the task is finished rather than stopping after one pass. A solo builder reaches for it when a job needs multiple rounds of effort and they want the agent to persist autonomously until done.
- Iterative agent task loop
- Repeats until task complete
- Progress-checked execution
Task Loop by the numbers
- 25 all-time installs (skills.sh)
- Ranked #9,800 of 16,546 AI & Agent Building skills by installs in the Skillselion catalog
- Data as of Aug 4, 2026 (Skillselion catalog sync)
npx skills add https://github.com/duyet/claude-plugins --skill task-loopAdd your badge
Show developers this skill is listed on Skillselion. Paste this into your README.
| Installs | 25 |
|---|---|
| repo stars | ★ 10 |
| Last updated | July 24, 2026 |
| Repository | duyet/claude-plugins ↗ |
What it does
Runs an iterative task loop where the agent repeatedly works, checks progress, and continues until a task is complete.
Who is it for?
Builders running multi-step autonomous agent tasks
Skip if: One-shot prompts
Files
Task Loop
Duyetbot's loop-based approach to task execution.
Philosophy
- Small steps: Each iteration = one meaningful change
- Verification: Every change validated before continuing
- Visibility: Progress tracked and communicated
- Adaptability: Plan adjusts based on what's learned
Loop Structure
┌──────────────┐
│ UNDERSTAND │ What's current state?
└──────┬───────┘
▼
┌──────────────┐
│ PLAN │ What's single next step?
└──────┬───────┘
▼
┌──────────────┐
│ EXECUTE │ One change only
└──────┬───────┘
▼
┌──────────────┐
│ VERIFY │ Did it work?
└──────┬───────┘
▼
Complete? ──NO──► Loop
│
YES
▼
DONEIteration Template
### Iteration N
**State**: What's done / pending
**Goal**: What this iteration accomplishes
**Execution**:
[1] Action → Result
**Verify**:
- [ ] Works as expected
- [ ] Tests pass
**Next**: What comes afterProgress Format
[x] Step 1: Done (iter 1)
[x] Step 2: Done (iter 2)
[ ] Step 3: Current (iter 3)
[ ] Step 4: PendingTermination
Success
- All acceptance criteria met
- Tests passing
- Code clean
Stop
- Blocker requiring human input
- Max iterations reached
- Same step failed 3x
Anti-Patterns
Don't: Giant Iterations
BAD: Iteration 1: Implement entire feature
GOOD: Iteration 1: Data model
Iteration 2: Core logic
Iteration 3: Error handling
Iteration 4: TestsDon't: Skip Verification
BAD: Execute → Execute → Execute → Check
GOOD: Execute → Verify → Execute → VerifyDon't: Ignore Failures
BAD: Test failed, moving on
GOOD: Test failed, investigating cause