finalise-plan
Finalise Plan
Pressure-test the plan, then repair it. Do not stop at critique if you can make the plan concrete and safer yourself.
Review the plan in context
- Read the user request, surrounding discussion, and current plan before changing anything.
- Identify the plan artifact to update. Prefer revising the existing plan in place instead of producing a separate review unless the user asked for findings only.
- If key context is missing and would materially change the plan, ask the user. Otherwise, record the assumption in the plan and continue.
If the workspace already has a standard plan structure, preserve it instead of inventing a new format.
Look for plan defects
Check for:
- missing scope boundaries, prerequisites, discovery work, or dependency checks
- incorrect sequencing or tasks that jump past unresolved blockers
- tasks that are too broad, mixed, or not independently verifiable
- acceptance criteria or validation steps that are vague, missing, or impossible to run
- hidden irreversible actions, rollout risks, or missing cleanup and follow-up work
- assumptions stated as facts
- places where ownership, ordering, or gating rules are still unclear
Repair the plan
Update the plan directly to fix what you found:
- add missing discovery, implementation, validation, rollout, or reporting steps
- reorder tasks to respect dependencies
- split broad tasks into atomic units
- make acceptance criteria and validation explicit
- document unresolved risks or assumptions when they cannot yet be removed
Enforce execution boundaries
For implementation plans, every task should be small enough to complete, validate, and commit independently.
Each implementation task should state:
Validation:what must pass before the task is doneCommit:the exact boundary that should be committed immediately after validation
If a task cannot support an independent validation and commit boundary, revise the plan into smaller tasks first.
For non-code plans, apply the same rule using the smallest independently verifiable completion boundary available.
Finish with a quality check
Before stopping, confirm the revised plan is:
- concrete instead of aspirational
- sequenced against real dependencies
- explicit about assumptions and open risks
- testable at the task level
- safe to execute without batching unrelated work
Then give a brief summary of the key corrections and any remaining open risks.
Default stance
- prefer fixing the plan over merely describing problems
- prefer the smallest plan changes that remove real risk
- prefer explicit assumptions over hidden ones
- prefer validation as part of the plan, not a postscript
- prefer atomic tasks over broad phases when execution details matter
- stop when the plan is concrete, sequenced, and testable
More from sebkay/skills
generate-agent-instructions
Generate or update AGENTS.md for AI coding agents by extracting project-specific architecture, workflows, conventions, and integration details from the repository. Use when the user asks to create, refresh, or improve AGENTS.md or agent instructions for a codebase.
23audit-dead-code
Find dead code and cleanup candidates such as unused exports, unreachable branches, orphaned files, stale feature flags, dead registrations, and compatibility layers with no live callers. Use when auditing refactors, bundle-size cleanup, architecture simplification, pre-release cleanup, reviewing requests to find unused code or decide what can be deleted, or when deciding whether code can be safely removed or auto-fixed.
20commit
Create a focused git commit for the unit of work just completed. Use when the user asks to commit changes, record finished implementation work, save a fix or refactor to git, or prepare a clean commit from the current branch with a concise message.
16ask-plan-questions
Ask high-impact clarifying questions before implementation to reduce mistakes, rework, and hidden assumptions. Use when creating, reviewing, or updating a plan; when requirements are ambiguous; or when scope/constraints are not fully specified.
15add-plan-todos
Create and maintain granular todos in the active execution plan to reduce omissions, sequencing errors, and misunderstandings. Use when drafting a plan, updating scope, starting implementation, or recovering from uncertainty/blockers.
12audit-errors
Audit code for error-handling inconsistencies, anti-patterns, and silent failures such as empty catch blocks, ignored promise rejections, log-and-continue paths, fallback values that hide faults, broad catch clauses, and inconsistent error translation across layers. Use when reviewing controllers, services, jobs, API handlers, async workflows, UI actions, or any change where errors may be swallowed, downgraded, or surfaced unreliably.
9