Use when a task needs single-agent self-correction across multiple iterations β write plan, execute, critique own output, revise plan, re-execute, until convergence or budget exhausted. Different from `agent-debate` (which is 2 agents arguing pro vs con); this is 1 agent looping
Installs just this skill. Get the whole plugin for auto-invocation.
β‘ How it fires
How this skill gets triggered: by you, by Claude, or both.
Fires itselfClaude auto-loads it when your prompt matches the work.
You can call itInvoke it directly when you want it.
Slash command/agent-plan-act-reflect
ποΈ Context preview
The summary Claude sees to decide when to auto-load this skill.
Use when a task needs single-agent self-correction across multiple iterations β write plan, execute, critique own output, revise plan, re-execute, until convergence or budget exhausted. Different from `agent-debate` (which is 2 agents arguing pro vs con); this is 1 agent looping
π Stats
Stars16
Forks4
LanguagePython
LicenseMIT
π¦ Ships with agent-collab-workspace
</> SKILL.md
agent-plan-act-reflect.SKILL.md
---name: agent-plan-act-reflect
description: Use when a task needs single-agent self-correction across multiple iterations β write plan, execute, critique own output, revise plan, re-execute, until convergence or budget exhausted. Different from `agent-debate` (which is 2 agents arguing pro vs con); this is 1 agent looping over its own work.
---# agent-plan-act-reflect
Single-agent iterative self-correction loop. Composes existing
building blocks (task-splitter / shared-memory / acceptance-gate)
into one closed cycle.
This is the **agent equivalent of test-driven development**: write
spec, build, test, fail, revise, build again, until pass. Where
`agent-debate` is "2 agents disagree", PAR is "1 agent revises".
## When to use
Good for:
- A single agent owns the task end-to-end, but the first attempt
rarely lands clean (refactors, papers, optimizations, anything
with > 3 quality dimensions to balance)
- The success criterion is automatically checkable (tests, eval, linter)
- Budget allows 2-5 iterations
- You want a paper trail of what was tried + why it failed
Not for:
- Single-shot tasks (overkill)
- Tasks where two agents should argue opposing designs (use `agent-debate`)
- Multi-agent coordination (use `agent-task-splitter`)
