plan-mode skill
✓ VerifiedEnforces a structured planning protocol that requires investigation, risk analysis, and explicit user approval before any code modification.
- GitHub stars
- 173
- Files
- 3
- Last commit
- Nov 11, 2026
- License
- MIT
- Allowed tools
- —
What does the plan-mode skill do?
plan-mode enforces a structured planning protocol that requires investigation, risk analysis, and explicit user approval before any code modification.
Summary generated from the skill's SKILL.md.
About the plan-mode skill
Developers who need to manage complex software changes can use this skill to enforce a strict planning protocol before any code is modified. It acts as a mandatory gate that requires an agent to investigate project context, analyze tasks, and surface risks or blockers before moving to implementation. By following a specific sequence of understanding, planning, and obtaining approval, it prevents premature or accidental mutations to the codebase.
An agent should load this skill when tasked with designing features, refactoring, migrations, or API changes. During the planning phase, the agent is restricted to read-only actions such as searching the repository, inspecting schemas, and reviewing tests. It is not intended for trivial one-line edits or simple questions about the code. The process concludes only after the user provides explicit approval for an evidence-based, step-by-step plan, ensuring that implementation only occurs once all ambiguities and dependencies are resolved.
What an agent can do with it
- 1Feature design and implementationAsk the agent to design and scope a new feature while ensuring all dependencies and risks are identified first.
- 2Safe codebase refactoringRequest a refactor of existing code to ensure the agent analyzes the current versus desired behavior before changing files.
- 3Controlled API changesDirect the agent to plan an API change by reviewing existing contracts and schemas without making immediate edits.
Written from the skill's SKILL.md. Every use case is checked against a quote from it.
How to install the plan-mode skill
Each command copies the skill folder into the client's skills directory. Restart the client or start a new session to load it.Install the plan-mode skill in Plugin
/plugin marketplace add CodeAlive-AI/ai-driven-development /plugin install ai-driven-development@ai-driven-development
Install the plan-mode skill in Claude Code
~/.claude/skills/plan-mode/SKILL.md
$ npx degit CodeAlive-AI/ai-driven-development/skills/plan-mode ~/.claude/skills/plan-modeInstall the plan-mode skill in Claude Code (project)
.claude/skills/plan-mode/SKILL.md
$ npx degit CodeAlive-AI/ai-driven-development/skills/plan-mode .claude/skills/plan-modeInstall the plan-mode skill in Codex
~/.agents/skills/plan-mode/SKILL.md
$ npx degit CodeAlive-AI/ai-driven-development/skills/plan-mode ~/.agents/skills/plan-modeInstall the plan-mode skill in Gemini CLI
~/.gemini/skills/plan-mode/SKILL.md
$ npx degit CodeAlive-AI/ai-driven-development/skills/plan-mode ~/.gemini/skills/plan-modeInstall the plan-mode skill in Claude.ai
npx degit CodeAlive-AI/ai-driven-development/skills/plan-mode plan-mode && zip -r plan-mode.zip plan-mode # claude.ai → Customize → Skills → + → Create skill → Upload a skill → plan-mode.zip
Automated checks
Ran Nov 14, 2026Full report →Passed our automated checks. That is not a security audit: read the scripts before you let an agent run them. How checks work
- Source repository linked and reachablecodealive-ai/ai-driven-development/skills/plan-mode is public and reachablegithub.com/codealive-ai/ai-driven-development/tree/HEAD/skil ↗Pass
- Open source license (OSI)MITspdx.org/licenses/MIT.html ↗Pass
- Commit or release in the last yearLast commit 2026-11-11github.com/codealive-ai/ai-driven-development/commits ↗Pass
- No download-and-run or credential theft in scriptsNo scripts in the skill folderPass
- No hidden instructions in SKILL.mdNo override phrases, safeguard switches, credential requests or hidden charactersgithub.com/codealive-ai/ai-driven-development/blob/HEAD/skil ↗Pass
- Pre-approved tools are scopedNo pre-approved tools; the agent asks before using onegithub.com/codealive-ai/ai-driven-development/blob/HEAD/skil ↗Pass
SKILL.md
View on GitHubPlan Mode
Prevent premature implementation. Understand → plan → get approval → implement → validate. Do not skip the gate just because a change looks small.
The full protocol — investigation steps, issue classification, the question and plan templates, validation, and special cases — lives in PROTOCOL.md. Read it before planning or implementing.
Default behavior
- Investigate project context (read-only).
- Analyze the task: goal, current vs desired behavior, acceptance criteria, scope, edge cases.
- Classify open issues as ambiguity / contradiction / blocker / risk / assumption.
- Ask the smallest set of focused questions that unblocks planning.
- Produce a consistent, evidence-based, step-by-step plan.
- Wait for explicit approval.
- Implement only the approved plan.
- Validate, then report what was and was not checked.
The gate (core rule)
During Plan Mode, do not mutate the project: no editing files, scaffolding, patches, package installs, lockfile updates, migrations, deploys, destructive scripts, or "trying an implementation" to learn the code.
Allowed: read files, search the repo, inspect structure, review tests / docs / configs / API contracts / schemas / migrations, run read-only or non-mutating diagnostic commands, and consult external docs when project context is insufficient.
If a tool or command might mutate state, ask first or avoid it.
Surfacing blockers
Surface ambiguities, contradictions, and blockers before the plan — never bury them inside it. For each, capture: what it is, the evidence, why it matters, whether it blocks implementation, and a proposed default or resolution. If none exist, state No hard blockers found.
Asking questions
Ask only after investigating context, and never for what the repo, tests, docs, or provided context already answer. Group related questions, give each a recommended default, and explain briefly why each answer matters. Avoid open-ended "What should I do?" questions.
Approval
Counts as approval: an explicit "proceed / implement / apply / go ahead / approve", or an edited plan with a clear instruction to continue. Does not count: answering one question, commenting on the plan, asking for more detail or alternatives, or "is this enough?". When approval is ambiguous, ask for a direct confirmation.
If new evidence contradicts the approved plan mid-implementation, pause and return to planning with what changed, the evidence, the impact, and a revised recommendation.
Reference
Full protocol, output templates, and special handling (small tasks, urgent fixes, user-provided plans, partial context): PROTOCOL.md.
Files in this skill
3plan-mode skill questions
How do I install the plan-mode skill in Claude Code?
Run npx degit CodeAlive-AI/ai-driven-development/skills/plan-mode ~/.claude/skills/plan-mode. Claude Code loads skills from ~/.claude/skills (all projects) or .claude/skills (one project) and uses a skill when its description matches the task.
Which clients can use the plan-mode skill?
Any client that reads the Agent Skills format: Claude Code, Codex, Gemini CLI and claude.ai (upload as a ZIP). Scripts in the skill run only where the client allows code execution.
How does the agent handle blockers and ambiguities?
The agent must surface ambiguities, contradictions, and blockers before creating the plan, detailing the evidence, impact, and a proposed resolution for each.
What constitutes user approval to proceed?
Approval is granted through explicit commands like 'proceed', 'implement', or 'go ahead', or by an edited plan with instructions to continue.