How to use the 90-Day Execution Plan agent skill (execute)

What does the execute skill make the agent do?

Given an approved decision, the agent breaks the chosen option into 3-6 workstreams and assigns a responsible person (DRI) to each. It then reverse-engineers 12 weekly milestones from a day-90 checkpoint date, sets a cadence of weekly and bi-weekly reviews plus day 30, 60 and 90 checkpoints, and builds a risk register that cross-references concerns raised earlier in the decision process.

The result is saved as a markdown file in ~/.claude/execution/, named with the date and a slug. The skill structures the plan; it does not carry out any of the work itself.

• Workstreams with DRI, success metric and status
• Weekly milestones with a definition of done for each week
• Cadence: weekly owner review (15 min), bi-weekly cross-functional sync (30 min), checkpoints at day 30, 60 and 90
• Risk register with likelihood, impact, owner and mitigation
• Kill criteria watch, copied from the decision and reviewed at every checkpoint

How do you install the execute skill?

Download the ZIP from this page and unzip it. For a personal install, place the skill folder in ~/.claude/skills/. To scope it to one project, place it in that project's .claude/skills/ folder.

If you use the Claude apps instead, upload the skill under Settings → Capabilities → Skills.

• Personal: ~/.claude/skills/
• Single project: .claude/skills/ inside the project folder
• Claude apps: Settings → Capabilities → Skills, then upload

What do you need before you run it?

The skill requires Claude Code or another SKILL.md-compatible agent, with the c-level-agents plugin installed. Its input must be an approved decision record, which is the output of /c-level-agents:decide. Without that record, the plan would rest on choices that have not been settled.

• Claude Code or a SKILL.md-compatible agent
• The c-level-agents plugin installed
• An approved decision record produced by /c-level-agents:decide

How do you trigger the skill with a natural request?

Point the agent at a decision record and say what you want. The command form is /c-level-agents:execute followed by the path to the decision. A natural request works too, as long as it names the approved decision and asks for an operating plan. Mention the date you want to hit the checkpoint if it matters, since the milestones are counted back from it.

• Turn the approved market-entry decision into a 90-day plan with weekly milestones and DRIs.
• Create an execution plan from the decision in decisions/pricing-change.md, with a checkpoint at the end of the quarter.
• Build the weekly milestones and risk register for the approved hiring decision.

What does a finished plan look like?

The plan follows a fixed template. The header records the linked decision, the sponsor, the start date and the checkpoint date. The outcome section copies the success and kill criteria from the decision and marks them binding. The remaining sections hold the workstream table, the week-by-week milestone table, cadence, dependencies, the risk register and the kill criteria watch.

Review the DRI column first. Each name should be a person who has agreed to the role, and each milestone should have an observable definition of done.

• Header: decision link, sponsor, start date, checkpoint date
• Outcome (binding): success and kill criteria
• Workstreams, weekly milestones, cadence, dependencies, risk register, kill criteria watch

What are the safety notes?

According to the source, this skill runs no scripts and uses no network access. It reads the decision record and writes a markdown file. Check the saved plan before sharing it, because it names people and may contain sensitive business details from the decision.

Do not assign DRIs who have not accepted the role. The plan will assign work to them on paper, and the skill cannot confirm their agreement.

• Runs scripts: no
• Uses network: no
• Output is a local markdown file; review names and details before sharing

Tips for getting a useful plan

Start from a decision that is formally approved and recorded. Keep the workstreams tied to the chosen option. If a checkpoint shows the plan is wrong, the source suggests re-deciding through /c-level-agents:boardroom rather than quietly editing the plan.

• Use the full workflow: office-hours, brief, boardroom, decide, then execute
• After day 90, or earlier if kill criteria trigger, run /c-level-agents:post-mortem
• Use the linked skills coo-advisor, strategic-alignment and change-management for background