Use when you have a written implementation plan to execute inline in this session, with checkpoints - the non-fleet execution path.
Scanned 9/3/2026
Install to Claude Code
npx -y skills add andreymudri/claude-teammates --skill executing-plans --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Executing Plans?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/andreymudri-executing-plans)More formats (shields.io, HTML) on the badges page.
---
name: executing-plans
description: Use when you have a written implementation plan to execute inline in this session, with checkpoints - the non-fleet execution path.
---
# Executing Plans
_Adapted from the MIT-licensed superpowers plugin by Jesse Vincent. See NOTICE.md._
Execute a written plan inline, in this session, task by task, checkpointing with your human
partner as you go. This is the counterpart to `parallel-execution`: same run state, different
cadence. Use `using-teammates` to decide which one applies before reaching for this skill.
## 1. Initialize the run
node "$CLAUDE_PLUGIN_ROOT/scripts/cli.mjs" init-run <planPath> --run <runId> --root <project root>
This writes `.teammates/<runId>/plan.json` and `status.json` — the same shared state a fleet
run would write. Because of that:
- `node "$CLAUDE_PLUGIN_ROOT/scripts/cli.mjs" digest --run <runId> --root <project root>`
renders this run unchanged, whether it ran inline or across teammates.
- `fleet-lifecycle resume <runId>` can resume it later.
- If the work turns out to be bigger than expected mid-run, hand it to `parallel-execution`
without translating anything — the state is already in the shape it expects.
Never start implementation on `main`/`master` without explicit consent from your human partner.
## 2. Review the plan critically
Read it fully before touching code. Raise concerns with your human partner before starting —
gaps, unclear steps, or a task that looks bigger than a single sitting.
## 3. Execute and checkpoint
Work through tasks in dependency order. Follow each step; run the verifications it specifies.
Checkpoint with your human partner **per task, or per small batch of related tasks** — not
automatically per phase the way a fleet gate does, and not silently at the end of the whole
plan. Report what changed, what you verified, and any deviation from the plan before moving on.
### Gate manifest is optional here
A fleet phase demands a `teammates.gate.json` manifest before it will pass. Inline work does
not need one. When no manifest exists, checkpoint against the plan's own verification steps
instead of demanding one be written first.
**Do not require a phase structure, a manifest, or a reviewer dispatch for small inline work.**
Ceremony that makes a two-task change expensive is ceremony that gets skipped — and a skipped
skill protects nothing. Reserve that machinery for work that has actually grown into a fleet
(see `parallel-execution` and `phase-gate`).
## 4. Finish
After all tasks are complete and verified, use `finishing-a-development-branch` (if available)
or otherwise confirm tests pass and present integration options to your human partner.
## When to stop and ask for help
Stop immediately when you hit a blocker (missing dependency, failing test, unclear
instruction), when the plan has a gap that prevents starting, or when a verification fails
repeatedly. Ask rather than guess.
## Remember
- Shared state with the fleet path: `.teammates/<run-id>/`, `init-run`, `digest`, resumable via
`fleet-lifecycle`.
- Checkpoint per task or per batch, with the user — not per phase, not automated.
- No gate manifest required; the plan's own verification steps stand in for one.
- Scale up to `parallel-execution` any time without rewriting state.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!