Skip to content
Agentmart
Submit a tool

implement-scoped

Scope requested work to fit one agent session, build what fits, finish the mechanical leftovers with sub-agents, and hand off the rest.

Skill by nash403

IndexedMIT
Write a review

Chalk Labs

Need help deploying implement-scoped?

Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.

Book a call

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/nash403-implement-scoped

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

Scope requested work to fit one agent session, build what fits, finish the mechanical leftovers with sub-agents, and hand off the rest. Use when a request's size is unclear or likely too big for one session.

Use cases: backend, 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/implement-scoped

What's inside

Skill name
implement-scoped
Description
Scope requested work to fit one agent session, build what fits, finish the mechanical leftovers with sub-agents, and hand off the rest. Use when a request's size is unclear or likely too big for one session.
License
MIT
Instructions
52 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

Scope the work to fit one session, build what fits with the usual implementation workflow (`implementation-loop`), and hand off the rest.

## 1. Gather

Dispatch a sub-agent to gather the facts, so the main context stays light for the build: the request, the ticket (from the prompt or the branch), and the codebase surface (Glob → Grep → Read only matches). It returns the files involved and the size signals below, not their content. Never ask what can be looked up.

## 2. Grill the gaps

Only if the facts leave the size undetermined, using `grilling` for the mechanism. Ask only what moves the estimate: capability count, back and/or front, new model or migration, impacted surfaces, what is explicitly out. UX detail and edge cases belong to the build.

## 3. Size

Judge, no threshold. High-weight signals: ≥2 distinct capabilities, backend plus frontend, new model, migration or schema, new external integration. Medium: new route or page, ≥3 files across ≥2 packages.

Anchors: **L** = model, migration, endpoints, UI. **M** = multi-file frontend with backend wiring. **S** = one surface, existing plumbing.

**At L or above, propose at least one fast path or defer.**

## 4. Fast paths and defers

Prefer a cheaper way to deliver the request over cutting it. Valid only when reversal is mechanical, needing no design-system or domain knowledge to finish later.

- **Defer**: capability leaves the session, stays visible as deferred (disabled control plus tooltip, banner). Never an enabled control that does nothing.
- **Cheap mechanics**: capability works, built cheaply. Raw strings over i18n keys, mocked data in a schema-compatible shape, placeholder loading, error and empty states.
- **Never**: native elements over the project's components, hand-rolled date manipulation or formatting when the project already offers them, hardcoded colors, skipped auth checks.

## 5. Gate

Cutting a capability is my call. When the scope defers one, present ≤15 lines: scope (one sentence per capability), 2-4 acceptance criteria, defers and fast paths with a one-line rationale each, out of scope, assumptions. Wait for an explicit go. I may amend the tradeoffs: apply and re-present once. A second round, or a shifted scope, means re-grilling.

Without a defer, state the scope and the fast paths in a few lines and go straight to the build.

## 6. Build

Build the approved scope only, following `implementation-loop` up to its review. Capabilities surfacing mid-build are defers.

## 7. Finish

Once the build passes its checks, take the fast paths that can be reversed now: mechanical, and nothing they need is missing (mocks stay mocks until the real API exists). Dispatch sub-agents to reverse them, one per set of files no other touches: usually a single one, since such fixes often share a file (a translations file). Brief each with the exact items, their files and how each reverses, and have it keep the checks green. A fix that fails, or turns out to need a decision, goes back to the handoff.

Then finish `implementation-loop` on the whole change, build and finishing together: review, then commit as it says.

## 8. Close and hand off

Restate what was deferred and why, then `handoff` the remainder:

- One handoff per substantial defer, the capability as argument.
- One shared handoff for small defers and the fast paths still in place, each listed with what reverses it (raw strings to i18n keys, mocks to real data, placeholder states to final ones).

List the handoff paths in the close.

From nash403/skills, MIT.

More from nash403/skills

Signals

Stars
0
Forks
0
Open issues
0
Last commit (this folder)
Oct 21, 2026 (10 days ago)
Last release
Oct 22, 2026 (8 days ago)

From GitHub (fetched Oct 24, 2026).

Details

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

More from nash403

Done for you

Rather not set it up yourself?

Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.

See how we build AI agents →Read the case studies

Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.

launches shipped
50+
launches shipped
media placements
500+
media placements
founders on every account
2
founders on every account

Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block

Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.

Get implement-scoped running in your business

Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying implement-scoped?

CallGet help
Cite this page

implement-scoped skill. Agent Mart. https://agentmart.dev/skills/nash403-implement-scoped (last updated 2026-10-24)