Author And Verify Implementation Plan logo

Author And Verify Implementation Plan

Organization
DasDigitaleMomentum
author-and-verify-implementation-plan

Author and verify per-phase implementation plans by cross-checking the gated phase intent against the current source code and existing docs inventories.

Overview

PublisherDasDigitaleMomentum
Repositoryopencode-processing-skills
Skill nameauthor-and-verify-implementation-plan
Stars
61
Forks
3
Bundled files
2
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.

  • 2 bundled files

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

  • Open source

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

Installation

Install the Author And Verify Implementation Plan 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/DasDigitaleMomentum/opencode-processing-skills/tree/main/skills/author-and-verify-implementation-plan
  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/DasDigitaleMomentum/opencode-processing-skills.git /tmp/opencode-processing-skills
mkdir -p .claude/skills
cp -r /tmp/opencode-processing-skills/skills/author-and-verify-implementation-plan .claude/skills/author-and-verify-implementation-plan
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Author And Verify Implementation Plan 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 Author And Verify Implementation Plan 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 Author And Verify Implementation Plan 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.

Skill: Author and Verify Implementation Plan

This skill standardizes the second pass between planning and execution:

  • It authors or refines plans/<plan>/implementation/phase-N-impl.md
  • It verifies the plan against the current source code (and uses docs/** inventories when present)

This skill exists to prevent “hand-wavy” implementation plans and reduce drift between phase intent and repo reality.


When to Use

Use this skill when:

  • The phase is already gated (scope/DoD decided in plan.md + phases/phase-N.md).
  • You want the “how” (implementation/phase-N-impl.md) to be grounded in current code.
  • You plan to delegate execution later via execute-work-package.

Do not use this skill to:

  • Change the gated phase intent/scope (that is primary work via update-plan).
  • Generate documentation (use generate-docs / update-docs).

Execution Model

Roles

  • Primary (maintainer)

    • Owns phase gating and decisions.
    • Delegates authoring/verification of the implementation plan when helpful.
    • Reviews the resulting implementation plan for scope compliance.
  • Subagent (delegate or justified model alias)

    • Uses one fresh Delegate session for exactly one phase implementation plan.
    • Reads: phase + existing implementation plan (if any) + relevant docs + relevant code.
    • Writes: plans/**/implementation/phase-N-impl.md.

Routing Matrix (Who does what)

  • Writes: plans/<plan>/implementation/phase-N-impl.md (authoring/refinement grounded against current code).
  • Does NOT write: plans/<plan>/plan.md or plans/<plan>/phases/** (phase intent/scope/DoD remain gated by the primary).
  • Primary: owns phase gating; reviews impl-plan for scope compliance.
  • delegate: performs the author+verify pass and writes the impl-plan. Use delegate-strong only when phase complexity or risk justifies it.
  • implementer: not used for plan artifacts in this skill.

Why plans/ and docs/ matter

  • plans/ is the gated source of truth for intent/scope/DoD.
  • docs/ (if present) provides curated inventories (modules/features/symbols) to navigate quickly.

Scope and Specification Boundary

Author the smallest complete implementation plan: the phase outcome must work, affected real paths must integrate, and applicable existing invariants must be preserved. Do not invent product rules, guardrails, configurability, abstractions, or future-proofing. Record a genuine user-owned fork as a blocking Reality Check for the Maintainer; resolve codebase-answerable questions yourself. Runtime values that operators may change belong in the project's existing configuration pattern.

The full boundary (gold-plating definition, configurable-value rule, and blocking-decision handling) is owned by the execute-work-package skill; this summary is not a separate policy.


Workflow

0) Inputs (Primary → delegate)

Provide:

  • Plan references:
    • plans/<plan>/plan.md (recommended: provides global constraints, scope language, and phase table)
    • plans/<plan>/phases/phase-N.md (required: gated intent/DoD for the target phase)
    • plans/<plan>/implementation/phase-N-impl.md (if exists)
    • plans/<plan>/implementation/phase-(N-1)-impl.md (optional: when the phase builds on previous technical decisions)
  • Only relevant existing docs references, if they materially reduce code discovery
  • Any constraints that must be preserved (naming conventions, verify command preference, etc.).

1) Author + verify

Primary starts a fresh canonical delegate session for the target phase using the bundled prompt template. A justified model alias may be selected without changing the workflow. This is a skill-defined artifact with an explicit output path and canonical template, so it does not require an informal Blueprint.

The delegate:

  1. Reads the plan (for global context) and the phase intent/DoD.
  2. Locates the relevant code areas using docs inventories and targeted code search.
  3. Designs the smallest complete technical change, reusing current structures unless a new element is presently necessary.
  4. Performs one deletion pass: removes or merges any step, new artifact, abstraction, or infrastructure that can be omitted without violating gated scope or a concrete existing invariant. This is author-owned QA, not a review loop.
  5. Writes/updates the implementation plan using the template and ensures it is concrete:
    • references real file paths/symbols
    • traces every implementation step to an authorizing requirement, scope item, acceptance criterion, or existing invariant that must be preserved
    • explains present necessity in the step's Why, rather than adding future-oriented work
    • justifies new modules, layers, interfaces, shared utilities, migrations, or infrastructure with evidence that a direct change to existing structures is insufficient
    • includes a single proposed verify command (or preserves the given one)
    • captures only material mismatches as “Reality Check” notes
    • marks any necessary but ungated user-owned decision as blocking and does not plan work that depends on it

If the phase depends on previous phases, its fresh delegate reads the necessary completed implementation plans to keep continuity. The Maintainer retains phase ordering and cross-phase decisions.

2) Primary review

Primary confirms:

  • The implementation plan stays within the gated phase scope.
  • No evident step or new structure is unnecessary for the phase.
  • The verify command is appropriate.
  • The “Reality Check” section (if any) is acceptable.

Output Contract

The generated/updated plans/<plan>/implementation/phase-N-impl.md MUST include:

  • A Required Context section with specific files.
  • Implementation Steps that reference concrete targets (files/symbols/components).
  • An Authorized By reference for every implementation step.
  • A single verify command in the Testing Plan that exercises changed behavior.
  • A Reality Check section only when material mismatches or blocking decisions exist.

The implementation plan must follow the canonical frontmatter and required headings from the bundled template. Omit sections marked optional when irrelevant.


Rules

  • Skill-first: when invoked, follow this workflow and template.
  • Fresh phase sessions, sequential processing. When authoring implementation plans for multiple phases, the Maintainer starts one fresh Delegate session per phase and processes phases strictly one at a time in dependency order. Each delegate writes only its target phase plan; do not reuse the authoring task_id for another phase or author phases in parallel.
  • Artifact-based cross-phase continuity. Each later phase delegate reads the completed prior implementation plans and checks shared interfaces, naming, data-flow assumptions, and dependency ordering while authoring its own target. It reports any inconsistency that requires changing a prior artifact or gated decision to the Maintainer rather than silently taking ownership of another phase.
  • Independent review handoff. After all phase plans are authored, hand the ordered set to one fresh reviewer session by default. Do not create a consolidated artifact.
  • Do not change phase scope/DoD; record mismatches under "Reality Check" and raise to the primary.
  • Produce the smallest complete solution, not the smallest edit at the expense of correctness. Cover necessary real paths and preserve necessary error handling, tests, and existing invariants while rejecting speculative completeness.
  • Prefer direct changes and existing structures. Justify new modules, layers, interfaces, shared utilities, migrations, or infrastructure in the relevant step's Why.
  • Do not generalize current work for hypothetical later phases. Preserve an already-gated cross-phase interface, but do not pre-implement or future-proof behavior.
  • Unspecified product, policy, or operational behavior is not authorization to add rules or guardrails. Preserve applicable existing invariants and avoid concrete regressions, but do not invent policy or exhaustively plan hypothetical edge cases.
  • If a necessary user-owned decision is not gated, record the exact decision as a blocking Reality Check item for the Maintainer and stop before planning dependent work. Resolve codebase-answerable questions and select purely local, reversible technical details when they do not change observable behavior.
  • Omit testing detail beyond the primary verify command, rollback, edge-case, deployment, and documentation planning unless explicit scope or concrete risk needs it.
  • Perform exactly one author-owned minimality pass before returning. Do not start an author/review/rewrite loop.

Templates

  • tpl-implementation-plan.md — canonical implementation plan format with “Reality Check” grounding section
  • tpl-author-and-verify-implementation-plan-prompt.md — Primary → delegate delegation prompt

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 Author And Verify Implementation Plan AI skill do?

Author and verify per-phase implementation plans by cross-checking the gated phase intent against the current source code and existing docs inventories.

Why use Author And Verify Implementation Plan on TypingMind?

Because you install it once and use it with any model. Author And Verify Implementation Plan 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 Author And Verify Implementation Plan in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/DasDigitaleMomentum/opencode-processing-skills/tree/main/skills/author-and-verify-implementation-plan. 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 Author And Verify Implementation Plan?

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 Author And Verify Implementation Plan?

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

Is the Author And Verify Implementation Plan AI skill free?

Yes. It is published on GitHub by DasDigitaleMomentum 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 👇