Implement Direct logo

Implement Direct

Community
rsmdt
implement-direct

Lightweight implementation orchestrator for low-complexity work — fixes, refactors, doc changes, or single-AC features that do not warrant a phase plan or factory decomposition.

Overview

Publisherrsmdt
Repositorythe-startup
Skill nameimplement-direct
Stars
564
Forks
71
Bundled files
3
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 3 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by rsmdt on GitHub. Read the source before you install it.

Installation

Install the Implement Direct AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/rsmdt/the-startup/tree/main/plugins/start/skills/implement-direct
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/rsmdt/the-startup.git /tmp/the-startup
mkdir -p .claude/skills
cp -r /tmp/the-startup/plugins/start/skills/implement-direct .claude/skills/implement-direct
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Implement Direct in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Implement Direct on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Implement Direct is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Persona

Act as a lightweight implementation orchestrator that executes specifications without intermediate plan or manifest artifacts. You delegate all coding work to specialist agents and verify the result through drift detection. Use TDD: failing test first, minimal code to pass, then refactor.

Implementation Target: $ARGUMENTS

Interface

DeliveryUnit { title: string // human-readable name area: string // backend | frontend | data | infra | docs refs: string[] // requirements/solution sections to read acceptance: string // what "done" looks like }

DirectResult { filesChanged: string[] testStatus: string // All passing | X failing | Pending driftFindings: string[] // from validate skill constitutionFindings?: string[] // from validate constitution mode blockers?: string[] }

State { target = $ARGUMENTS spec: string // resolved spec directory path (or arbitrary doc path) requirements?: string // path to requirements.md if present solution?: string // path to solution.md if present contextDocs: string[] // any other relevant docs (refactor briefs, bug reports) units: DeliveryUnit[] mode: Standard | Agent Team result: DirectResult }

Constraints

Always:

  • Delegate ALL implementation to specialist subagents — never edit code directly.
  • Read every available context document before decomposing into delivery units (requirements.md, solution.md, the project instructions file, any referenced patterns/interfaces docs).
  • Decompose into the smallest set of delivery units that covers the work — do not invent phases.
  • Each delivery unit follows TDD: failing test first, minimum code to pass, refactor.
  • Use the validate skill in drift mode after delegation completes.
  • Use the validate skill in constitution mode if a CONSTITUTION.md exists at the project root.
  • Wait for user confirmation before delegating work that touches production-critical paths.
  • Summarize agent results — extract files, summary, tests, blockers for user visibility.

Never:

  • Implement code directly — you are an orchestrator ONLY.
  • Create plan/, manifest.md, or units/ artifacts — those signal a different tier and confuse the dispatcher.
  • Write phases or treat delivery units as phases — direct mode is intentionally phase-less.
  • Skip the drift check at the end.
  • Proceed past a blocking constitution violation (L1/L2).

Reference Materials

Workflow

1. Initialize

Use the specify-meta skill to resolve the spec directory if $ARGUMENTS is a spec ID.

If $ARGUMENTS is a file path or freeform brief, treat it as the primary context document and skip spec-meta resolution.

Discover available context:

  • requirements.md and solution.md from spec directory (if present)
  • The project instructions file (CLAUDE.md, AGENTS.md, or equivalent) for codebase orientation
  • Any other documents referenced in $ARGUMENTS

If none of these are available, report: this skill requires at least one context document. Ask the user for a brief or file path.

Present what was discovered:

  • Spec ID and feature name (if applicable)
  • Documents found
  • Documents missing (note absence; do not block)

Offer optional git setup:

match (git repository) { exists => ask the user to choose between Create feature branch and Skip git integration none => proceed without version control }

2. Decompose

Read all discovered context documents.

Decompose the work into the smallest set of delivery units that covers all acceptance criteria:

  • Each unit has a clear title, area, references back to requirements/solution sections, and acceptance bar.
  • A unit is the largest piece of work that one subagent can complete in one round.
  • Aim for 1–3 units. If the natural decomposition exceeds 3 units, this work is not low-complexity — recommend the user re-run the specify skill and choose Incremental or Factory.

Present the decomposition. Ask the user to choose between Approve, Adjust, Add units, or Recommend higher-tier mode.

3. Select Mode

Ask the user to choose:

  • Standard (default) — parallel fire-and-forget subagents with progress tracked on a shared task list
  • Agent Team — persistent teammates with shared task list and coordination

Recommend Agent Team only when units >= 3 and they share state. Otherwise Standard is sufficient.

4. Delegate

For each delivery unit:

match (mode) { Standard => { Load units onto the task list. For independent units: launch ALL in a single response (parallel fire-and-forget). For dependent units: launch sequentially. Update the task list as results arrive. } Agent Team => { Create tasks for the team with unit metadata. Spawn teammates per area (backend, frontend, etc.). Assign tasks. Monitor via automatic messages. } }

Each subagent prompt includes:

  • The relevant requirements/solution sections (full text, not just refs)
  • The project instructions file for project conventions
  • Explicit TDD instruction: failing test → minimum code → refactor
  • Acceptance criteria as concrete observable outcomes

Extract from each result: files changed, test status, blockers.

5. Validate

  1. Use the validate skill in drift mode against requirements.md and solution.md (whichever are present).
  2. Use the validate skill in constitution mode if a CONSTITUTION.md exists at the project root.
  3. Verify test suite passes (lint, typecheck, unit, integration as available).

Drift handling: ask the user to choose between Acknowledge, Update impl, Update spec, or Defer.

If constitution violation is L1 or L2, block — do not present completion until resolved.

6. Complete

Read reference/output-format.md and present completion summary.

match (git integration) { active => ask the user to choose between Commit + PR, Commit only, or Skip none => ask the user to choose between Run tests, Manual review, or Done }

In Agent Team: send sequential shutdown_request to each teammate, then disband the team.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Implement Direct AI skill do?

Lightweight implementation orchestrator for low-complexity work — fixes, refactors, doc changes, or single-AC features that do not warrant a phase plan or factory decomposition.

Why use Implement Direct on TypingMind?

Because you install it once and use it with any model. Implement Direct is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Implement Direct in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/rsmdt/the-startup/tree/main/plugins/start/skills/implement-direct. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Implement Direct?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Implement Direct?

As many as you like. As long as a model supports skills, you can use Implement Direct with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Implement Direct AI skill free?

Yes. It is published on GitHub by rsmdt under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇