Writing Plans

Break work into clear, ordered plans before execution. Use when a task is multi-step, long-running, easy to derail, or needs coordination across tools, files...

Security Scan

This is an instruction-only “planning” helper whose requirements, behavior, and scope are consistent with its stated purpose — it contains no code, no installs, and asks for no credentials.

Purpose & Capability

Name/description match the content: the skill provides structured guidance for breaking multi-step tasks into ordered plans. There are no unexpected binaries, env vars, or installs required.

Instruction Scope

The SKILL.md recommends actions such as 'inspect installed files and runtime requirements' and 'install missing low-risk dependencies.' Those are reasonable for a planning helper, but they imply the agent may read local files or attempt installs when following the plan — the skill itself does not declare or constrain those reads/installs.

Install Mechanism

No install spec or code files are present; this is instruction-only so nothing will be downloaded or written by the skill itself.

Credentials

The skill requests no environment variables, credentials, or config paths. Although it mentions possible sub-agents and external systems, it does not request tokens or secrets itself.

Persistence & Privilege

always is false and the skill is user-invocable. It does not request persistent presence or attempt to modify other skills or global agent settings.

Assessment

This skill is a generic planning guide and is internally consistent. Before following any plan the agent produces: review any step that reads local files or installs software, approve installs explicitly, avoid providing secrets or service credentials unless you expect them to be needed and trust the destination, and monitor/report the agent's actions when it interacts with external systems or runs tooling. If you want stricter limits, require human confirmation for file access or package installs triggered by plans.

Write plans that are concrete enough to execute, review, and verify.

Core Rule

Do not jump into a long task with only a vague idea. Break it into explicit steps with outcomes.

When to Use

Use for:

• multi-step setup work
• migrations
• skill installation and validation passes
• browser automation workflows
• refactors
• research + implementation tasks
• tasks that may require sub-agents

Plan Shape

A good plan includes:

1. Goal
2. Constraints
3. Ordered steps
4. Verification points
5. Possible blockers

Good Step Format

Each step should answer:

• what to do
• what tool/file it touches
• what counts as success

Example:

1. Inspect installed files and runtime requirements.
2. Install missing low-risk dependencies.
3. Run smoke tests.
4. Separate working / needs-auth / broken states.
5. Report results.

Execution Rule

After writing the plan:

• execute in order
• update the plan if reality changes
• do not hide changed assumptions

Lightweight vs Heavyweight

Lightweight plan

Use for 3-5 step tasks.

Heavyweight plan

Use when:

• many files or tools are involved
• external systems are involved
• the task may run long
• sub-agents may help

Reporting

Summarize with:

• planned scope
• what completed
• what changed during execution
• what remains

Practical Examples

Example: Skill setup pass

1. Inspect files and dependencies
2. Install missing low-risk runtime tools
3. Run smoke tests
4. Split results into working / needs-auth / blocked
5. Report next steps

Example: Browser automation task

1. Confirm target site and success condition
2. Test a minimal path first
3. Add login/session handling if needed
4. Re-test the actual target flow
5. Record blockers clearly