Skip to content
Agentmart
Submit a tool

delegation-plan

Write plans intended for a cheaper or less capable implementation model.

Skill by szokeasaurusrex

IndexedMIT★ 2
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/szokeasaurusrex-delegation-plan

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Write plans intended for a cheaper or less capable implementation model. Use when the user asks for a plan suitable for another LLM, junior engineer, intern, cheaper model, coding agent, or implementation delegate.

Use cases: ml, writing, agent-building

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/delegation-plan

What's inside

Skill name
delegation-plan
Description
Write plans intended for a cheaper or less capable implementation model. Use when the user asks for a plan suitable for another LLM, junior engineer, intern, cheaper model, coding agent, or implementation delegate.
License
MIT
Instructions
95 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Delegation Planning

Use this skill to write implementation plans for a capable but lower-reasoning executor.

## Goal

Write a plan that removes ambiguity without doing the implementation reasoning that the delegate can do cheaply.

Think like a senior engineer and prompt engineer:
- Specify the objective, constraints, and success criteria.
- Define clear boundaries and decision rules.
- Leave routine code navigation and implementation details to the delegate.

## Planning principles

- Do not paste full implementations unless the exact code is the requirement.
- Do not provide large code blocks that solve the task.
- Prefer descriptions of required behavior over exact replacement snippets.
- Give file/module targets, not line-by-line edits, unless the location is genuinely ambiguous.
- State what must not change.
- State how to handle known edge cases.
- Include validation commands and expected outcomes.
- Include escalation criteria: when the delegate should stop and ask.

## Required structure

Use this structure unless the user requests another format:

```markdown
# <Task title>

## Objective
<One paragraph describing the user-visible behavior to achieve.>

## Scope
- In scope: <files, modules, behavior>
- Out of scope: <explicit non-goals>

## Constraints
- <compatibility, minimality, performance, style, or API constraints>

## Implementation guidance
1. <Concrete step with target files/modules and intended behavior.>
2. <Concrete step.>
3. <Concrete step.>

## Edge cases
- <case>: <required handling>

## Tests
- <test to add/update and what it should prove>

## Validation
Run:
- `<command>`

Expected result:
- <what passing looks like>

## Stop and ask if
- <condition that requires human/senior decision>
```

## Specificity calibration

Use more detail for:
- Non-obvious invariants.
- Risky migrations.
- Security, data loss, auth, or compatibility concerns.
- Places where multiple valid implementations would conflict with user intent.

Use less detail for:
- Obvious syntax.
- Mechanical refactors.
- Test fixture boilerplate.
- Exact code the delegate can infer safely from surrounding code.

## Anti-patterns

Avoid:
- Writing exact helper functions when a behavior description is enough.
- Replacing the delegate's design work with full code.
- Hiding choices as assumptions.
- Saying "update tests" without naming the behavior each test must verify.
- Over-constraining harmless implementation details.

## Final self-check

Before finalizing, verify:
- A motivated intern could start without guessing the goal.
- The delegate still has meaningful implementation work to do.
- Every ambiguity that affects correctness has a rule or escalation path.
- The plan says how to prove the change works.

From szokeasaurusrex/pi-agent, MIT.

More from szokeasaurusrex/pi-agent

Signals

Stars
2
Forks
0
Open issues
0
Last commit (this folder)
Jul 6, 2026 (3 months ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from szokeasaurusrex

Cite this page

delegation-plan skill. Agent Mart. https://agentmart.dev/skills/szokeasaurusrex-delegation-plan (last updated 2026-10-24)