superpowers-executing-plans
Skill101 starsChanged 30 days agoMITView on GitHub
Executing a written plan yourself? Load first to track progress and verify each task.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- EvovexAI/EvoFlowtheir own version, first seen Oct 8, 2026
- karim-embarek/superpowers-antigravitytheir own version, first seen Aug 17, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- EvovexAI/EvoFlow★ 612same name
- karim-embarek/superpowers-antigravity★ 2same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository12 other files it gives its agents.
- AGENTS.mdAGENTS.md
- superpowers-brainstormingSkill
- superpowers-diagnosing-superpowersSkill
- superpowers-finishing-a-development-branchSkill
- superpowers-receiving-code-reviewSkill
- superpowers-sync-upstreamSkill
- superpowers-systematic-debuggingSkill
- superpowers-test-driven-developmentSkill
- superpowers-using-git-worktreesSkill
- superpowers-verification-before-completionSkill
- superpowers-writing-plansSkill
- superpowers-writing-skillsSkill
The file
We split this skill into the rules it gives an agent
9 rules. About 0 are common, 4 are distinctive, and 5 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 16
Verify an isolated worktree using superpowers-using-git-worktrees.
- Distinctive: line 17
Read the plan and linked spec.
- Very specific: line 18
Create a task list with Reasonix todo_write.
- Very specific: line 19
Create a ledger at ~/.reasonix/plan-ledgers/<repo>/<plan-name>.md (or another persistent, plan-specific location).
- Distinctive: line 22
Keep plan-specific artifacts together.
- Very specific: line 28
Read that task's exact brief and note BASE, the commit before its work.
- Very specific: line 29
Follow its steps in RED-GREEN order.
- Very specific: line 30
If code is wrong, load superpowers-systematic-debugging and find the cause.
- Distinctive: line 31
Commit as the plan directs.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Executing Plans
Execute the plan yourself, task by task, in this session. The plan carries the design; a persistent ledger carries your decisions and progress through compaction. TDD is the per-task gate, and a fresh whole-branch review is the final second opinion.
Announce at start: "I'm using the superpowers-executing-plans skill to implement this plan."
Continue from one task to the next without asking the user to say "continue." User instructions and approvals already given remain valid. Keep commentary concise; the ledger and test output carry the detail.
Setup
- Verify an isolated worktree using superpowers-using-git-worktrees. Do not implement on
mainwithout explicit authorization. - Read the plan and linked spec. If they conflict, the spec is binding; when neither answers, make a reasonable ruling.
- Create a task list with Reasonix
todo_write. Todos show current status; they do not replace the ledger. - Create a ledger at
~/.reasonix/plan-ledgers/<repo>/<plan-name>.md(or another persistent, plan-specific location). Its first line names the absolute plan path. Check for an existing ledger first; after compaction, trust its completed-task lines andgit lograther than your memory. Never overwrite a ledger for a different plan. - Load superpowers-test-driven-development before Task 1. Scan task interfaces for conflicts and record pre-flight findings in the ledger. Record
Pre-flight: no shared interfaceswhen none exist.
Keep plan-specific artifacts together. Don't remove another plan's ledger. If the ledger is lost, reconstruct completion from commits and tests before continuing.
Task Loop
For each incomplete task:
- Read that task's exact brief and note
BASE, the commit before its work. Mark the todo in progress. - Follow its steps in RED-GREEN order. Run each command and compare actual output with its
Expected:line. A test that passes before implementation needs investigation. - If code is wrong, load superpowers-systematic-debugging and find the cause. If the plan is wrong or ambiguous, make the smallest decision satisfying the spec and add
Task N: Ruling: <decision> — <why> — <cost if wrong>to the ledger. Continue. An unrecorded deviation is a hidden decision. - Commit as the plan directs. Verify every named test exists and ran, every expected output was checked, and the final task test command passed. Load superpowers-verification-before-completion before claiming task completion.
- Append
Task N: complete (commits <BASE>..<HEAD>, tests: <command> → <result>)to the ledger only after the test passes. Mark the todo complete and take the next task.
Record rulings and completion as work happens, not at the end. After compaction, resume at the first task without a completion line. A task spanning multiple commits uses its saved BASE, not HEAD~1.
When to Stop
Resolve routine conflicts and plan gaps by ruling and continuing. Stop only for an irreversible or destructive action, a security-sensitive action, an external side effect that normally needs user approval, or a plan so broken that every path forward is a guess. Existing user authorization satisfies a gate; do not ask twice.
Whole-Branch Review
After all tasks pass, review the full range from the branch's merge base to HEAD, not only the last commit. Give a fresh native Reasonix review or task reviewer the plan, spec, diff, ledger rulings, and the plan's Review Focus. If no independent reviewer is available, perform a separate self-review and disclose that limit.
Re-grade findings by the effect on a reasonable user, even when the spec is silent. Fix critical and important findings in one pass: reproduce each with a failing test, make it pass, run the full suite, and record Final: fixed <finding> — <test> RED→GREEN, suite <result>. Record deferred minor findings in the ledger and final handoff. If you decline a substantive finding, ledger a ruling with the cost if wrong. Do not silently discard reviewer concerns.
Then load superpowers-finishing-a-development-branch. Report the verification evidence and all rulings and deferred minors. Preserve the ledger until the branch result is settled; remove only this plan's scratch files when they are no longer needed.
Red Flags
| Thought | Reality |
| "I remember Task N" | Read the exact brief; memory is a summary. |
| "The test should pass" | Run it and read the output. |
| "I'll write ledger lines later" | Compaction may happen before later. Record each decision now. |
| "The plan is wrong, so I'll just do the right thing" | Rule, ledger the reason and cost, then continue. |
| "I should check in before the next task" | Continue through the authorized plan. |
| "I reviewed my own diff, so a fresh review adds nothing" | A fresh context can catch assumptions carried from implementation. |
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

