Core Methodology adds four skills to Claude Code. They change how Claude works before you ask it to: shaping an idea one question at a time, showing the decision path it took, refusing to call work finished without running a check, and naming the cause of a problem before changing anything.
01Install
Accept the GitHub invitation in your email. It gives your GitHub account read access to the plugin repository, and nothing else.
Until you accept it, the install command has nothing to fetch.
Run these three commands in Claude Code. No keys or extra software are required.
// 1. Register the Tenkr marketplace /plugin marketplace add TenkrAS/tenkr-plugin-marketplace // 2. Install the plugin /plugin install core-methodology@tenkr-plugin-marketplace
// 3. Wire it into the project you are working in
/core-methodology:setupThe @tenkr-plugin-marketplace suffix is required. It identifies the marketplace that provides the plugin.
Setup reads these seven things inside .claude/:
- Skills
- Agents
- Commands
- Hooks
- Settings
- Scripts
- Installed plugins
Setup writes one note to plugin memory. It changes nothing else.
If your tools overlap with the plugin's four skills, setup asks which should win. It records your answer.
Nothing is written into your project tree. An empty .claude/ is a clean install, not an error. Setup is safe to re-run.
02Confirm it works
Give Claude Code a small, real task you need today. Ask normally without mentioning the plugin.
Watch how the answer ends. Completion Proof blocks an unsupported finished claim.
A working result follows this sequence:
- Claude refuses to call the task finished.
- It names the check it has not run.
- It runs that check.
- It reports what the check showed.
That refusal means the skill is working.
A bare "done" with no checkable evidence is wrong. Watch for should, probably, and seems to attached to finished claims. Completion Proof treats each phrase as a stop signal.
03How the four skills work
Core Methodology adds four skills that change how Claude Code works before you ask it to. Each skill has its own trigger. You do not call the skills by name.
Idea Shaping
Idea Shaping runs before build or design work.
It first reads your source files, README, and specs. It does not ask questions your project already answers.
It asks one question at a time. It prefers two to four concrete options over an open-ended prompt.
The result is a design document, not code. It writes design-{slug}-{date}.md into the plugin workspace and stops.
You can disagree with the design before anything is built.
Structured Reasoning
Structured Reasoning divides a non-trivial answer into three parts:
| Part | Contents |
|---|---|
<analysis> | Restates the goal and constraints. |
<reasoning> | Weighs the options and explains why one won. |
<decision> | Contains the answer, with nothing after it. |
Read <decision> when you only need the result. Read the other parts when you want to challenge the reasoning.
When you push back, Claude rewrites <decision>. It does not append a correction below the original.
Completion Proof
Completion Proof requires evidence before Claude calls work finished.
- After updating data, Claude opens the result and checks the new values.
- After writing a document, Claude rereads it end to end and confirms the required sections.
- After fixing a problem, Claude reproduces the original problem and watches it not happen.
The check must run in the same message as the finished claim. An earlier result in the conversation no longer counts as evidence.
Root Cause Analysis
Root Cause Analysis blocks a fix until Claude names the cause.
Claude reads the full error, reproduces the problem, and lists what changed immediately before it appeared.
For a multi-step process, it finds the first step that turns good input into bad output. Later steps carry the error but do not cause it.
After two failed attempts, Claude stops guessing. It questions its understanding instead of trying a third patch.
04Use it day to day
There is nothing to run.
A primer loads at the start of every session. It lists the four skills and their triggers. This makes Claude use them consistently instead of occasionally.
Only Idea Shaping leaves files behind. It saves validated designs to workspace/core-methodology/, or the location you chose during setup.
Those files are yours to edit or delete. The other three skills produce only the answer in front of you.
/core-methodology:setup again so the four skills know about it. Nothing else in your project changes.05Troubleshooting
These two problems do not require a fresh purchase or a support ticket.
| What you see | Cause | Fix |
|---|---|---|
| The install command cannot find the plugin. | Your GitHub invitation is not accepted. Your account cannot see the plugin repository. | Accept the email invitation. Run the install command again. |
| Claude finishes tasks without evidence. | The session was already running when you installed the plugin. | Start a new Claude Code session. The skills and session-start primer load when a session opens. |
06Privacy, licence, and updates
Everything runs on your machine under your Claude subscription. Tenkr never sees your project, files, or prompts.
The plugin never calls a Tenkr service. There is no key to provide or service to sign into.
The plugin is a set of instructions loaded into your existing Claude Code installation.
It writes only the setup note and Idea Shaping designs. Both remain in locations you can read.
Core Methodology is MIT licensed. Your one-time purchase includes lifetime updates through the same install command.