Agent skill

Initializing Vertical Flywheel

by jianzhichun in jianzhichun/emerge

A skill your agent uses when a user asks to initialize a domain flywheel from natural language context, especially when environment details are incomplete or mixed with execution assumptions.

MITAuto-check passedTesting & QA

Install Initializing Vertical Flywheel

skills CLI
$ npx skills add jianzhichun/emerge --skill initializing-vertical-flywheel -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install jianzhichun/emerge initializing-vertical-flywheel --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/jianzhichun/emerge.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/initializing-vertical-flywheel .claude/skills/initializing-vertical-flywheel && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
initializing-vertical-flywheel
GitHub stars
106
Token cost
~2.2k tokens
SKILL.md length
923 words
Files
1
Skills in repo
14
Repo updated
First seen
Licence
MIT

At a glance

A skill your agent uses when a user asks to initialize a domain flywheel from natural language context, especially when environment details are incomplete or mixed with execution assumptions.

  • Works in 3 steps: RED: add/init tests for the requested… → GREEN: add minimum assets and code to… → REFACTOR: harden naming, verification,…
  • A user asks to initialize a domain flywheel from natural language context
  • SKILL.md covers Overview, When to Use, Mandatory TDD Flow and Core Initialization Contract, plus 11 more sections
  • Calls python3 and pytest

What it does

Initializing Vertical Flywheel is an agent skill from jianzhichun/emerge. Use when a user asks to initialize a domain flywheel from natural language context, especially when environment details are incomplete or mixed with execution assumptions.

Its SKILL.md is about 2.2k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Testing & QA. The licence is MIT.

When your agent uses it

  • A user asks to initialize a domain flywheel from natural language context
  • Especially when environment details are incomplete
  • Mixed with execution assumptions

Example prompts

  • “/initializing-vertical-flywheel”

Requirements

  • Python 3

Workflow steps

3 steps, taken from the first numbered list in SKILL.md.

  1. RED: add/init tests for the requested vertical and watch them fail.
  2. GREEN: add minimum assets and code to make tests pass.
  3. REFACTOR: harden naming, verification, and output shape without changing behavior.

What it can do on your machine

Read from SKILL.md and the folder at commit 035db30. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • python3
    • pytest

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Initializing Vertical Flywheel loads about 2.2k tokens when it runs. Until then it costs about 51 tokens; SKILL.md has 923 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~51
When it runs · the whole SKILL.md, loaded when a task matches
~2.2k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from jianzhichun/emerge at commit 035db30, republished under its MIT licence (© jianzhichun). 923 words, ~2,176 tokens.

Download SKILL.mdSave it as .claude/skills/initializing-vertical-flywheel/SKILL.md (or your agent's skills folder).
name
initializing-vertical-flywheel
description
Use when a user asks to initialize a domain flywheel from natural language context, especially when environment details are incomplete or mixed with execution assumptions.

Initializing Vertical Flywheel

Overview

Use this skill to convert one natural-language bootstrap request into concrete vertical flywheel assets under ~/.emerge/connectors/<vertical>/, plus verified runtime readiness. Never place vertical connectors inside the plugin project directory.

Core principle: do not claim initialization complete until new read/write pipelines execute and policy state is observable.

When to Use

  • User asks for one-sentence bootstrap, such as "initialize desktop_drafting_app vertical flywheel".
  • User context is natural language, not strict parameters.
  • Environment assumptions are uncertain (host/tooling/executor not guaranteed).

Do not use when:

  • User only asks for explanation, not initialization.
  • User asks only for policy status or read-only review.

Mandatory TDD Flow

  1. RED: add/init tests for the requested vertical and watch them fail.
  2. GREEN: add minimum assets and code to make tests pass.
  3. REFACTOR: harden naming, verification, and output shape without changing behavior.

No bootstrap completion claim without RED and GREEN evidence.

Core Initialization Contract

  • Input is user natural language; do not force CLI-style parameter declarations.
  • Extract only what is explicit in user text.
  • Ask only minimal clarifying questions when execution cannot proceed.
  • Produce:
    • bootstrap status (init_ok, degraded, or blocked)
    • created/updated assets
    • next verification actions

Remote runner setup (when needed)

If the user context indicates remote execution, initialize runner connectivity first.

Minimum steps:

  1. Generate install URLs from local plugin root (or Cockpit Monitors → Add Runner):
    • python3 "${CLAUDE_PLUGIN_ROOT}/scripts/repl_admin.py" runner-install-url --target-profile "<target_profile>" --pretty
  2. Send the printed command to the operator; they run it on the target machine (self-install).
  3. Verify with one icc_exec smoke call before creating vertical assets.
  4. Verify admin health signal:
    • python3 "${CLAUDE_PLUGIN_ROOT}/scripts/repl_admin.py" runner-status --pretty
    • proceed only when Runner reachable: True.

Runner HTTP protocol (summary):

EndpointPurpose
POST /runExecute one icc_exec call
GET /healthLiveness probe
GET /statusProcess info (pid, uptime, root)
GET /logs?n=NLast N log lines

Runner accepts only icc_exec. Pipeline bridge execution is handled by the daemon (loads files locally, sends as inline icc_exec). Request shape: {"tool_name": "icc_exec", "arguments": {"code": "...", "target_profile": "...", "no_replay": false}}.

Connector Location Rule

This plugin is a generic RWB flywheel engine. Vertical-specific connectors must NOT be placed inside the plugin project directory.

  • Plugin project (connectors/mock/) — testing only, committed to git
  • User verticals → ~/.emerge/connectors/<vertical>/ (user-space, not committed)
  • Override via env: EMERGE_CONNECTOR_ROOT=<path> (e.g. for remote runner deployments)

PipelineEngine searches ~/.emerge/connectors/ before the plugin root, so user verticals take precedence automatically.

Assets To Create (Minimum)

For vertical <vertical> (for example desktop_drafting_app), create in user-space:

  • ~/.emerge/connectors/<vertical>/pipelines/read/state.yaml
  • ~/.emerge/connectors/<vertical>/pipelines/read/state.py
  • ~/.emerge/connectors/<vertical>/pipelines/write/apply-change.yaml
  • ~/.emerge/connectors/<vertical>/pipelines/write/apply-change.py
  • ~/.emerge/connectors/<vertical>/synthesis_hints.yaml when reverse-flywheel synthesis should learn from operator events
  • tests (in plugin project, prefer existing suites unless there is a strong reason to split files):
    • tests/test_pipeline_engine.py
    • tests/test_mcp_tools_integration.py

Do not skip write verification hooks:

  • run_write(...)
  • verify_write(...)
  • rollback_write(...) when policy is rollback

Pipeline Metadata Rules

Each yaml/json metadata file must include:

  • intent_signature
  • *_steps (read_steps or write_steps)
  • verify_steps
  • rollback_or_stop_policy (stop or rollback)

Implementation Pattern

  1. Start with mock-safe behavior in *.py that returns deterministic objects.
  2. Ensure read returns structured rows and verify payload.
  3. Ensure write returns verification_state plus policy enforcement fields:
  • policy_enforced
  • stop_triggered
  • rollback_executed
  • rollback_result
  1. Keep output keys stable; do not introduce ad-hoc text-only outputs.

Verification Checklist

Run, in order:

  1. Targeted tests for new vertical files.
  2. pytest -q full suite.
  3. Confirm policy observability:
  • python3 "${CLAUDE_PLUGIN_ROOT}/scripts/repl_admin.py" policy-status --pretty
  • verify intent entry appears and counters move after calls.

Initialization is complete only when:

  • read and write icc_exec calls succeed (using the connector's intent_signature)
  • policy status includes the new intent key (shape: <connector>.<mode>.<pipeline>)
  • when remote mode is used, at least one call is confirmed through runner dispatch
  • tests and lint pass.

If runner is required and unreachable, return blocked (not degraded).

Show full SKILL.md (333 more words)Show less

TDD Test Surface (Required)

Do not treat PipelineEngine unit tests alone as sufficient proof.

For RED and GREEN phases, tests must exercise MCP-facing tool paths:

  • icc_exec (primary exploration tool — pipeline bridge execution is via icc_span_open when stable)

Minimum expectation:

  1. At least one failing-then-passing test through EmergeDaemon.call_tool(...) or JSON-RPC tools/call for each intent used by the init flow.
  2. At least one integration assertion that policy registry changes after icc_exec calls.
  3. If flywheel bridge is used, include a failing-then-passing test for bridge key updates (flywheel::...).

Quick Reference

  • Read pipeline id: <vertical>.read.<pipeline>
  • Write pipeline id: <vertical>.write.<pipeline>
  • Flywheel bridge key shape: flywheel::<pipeline_id>::<intent_signature>::<script_ref>
  • Policy states: explore -> canary -> stable

Rationalization Table

ExcuseReality
"We can assume remote-vm exists"Executor is optional context, not a guaranteed dependency.
"Mock connector means init is done"Init requires runnable assets, passing tests, and policy visibility.
"No need for TDD on docs/skills"Skills are process code; TDD still applies.
"I can ship only yaml metadata"Flywheel requires executable *.py and verification behavior.

Red Flags

  • "I can skip baseline and start implementation."
  • "I will hardcode environment assumptions from my local setup."
  • "I can declare success without verification output."
  • "I added files but did not run icc_exec integration tests."

Any red flag means stop and return to RED.

Reverse Flywheel Integration

When reverse-flywheel synthesis is in scope, write synthesis_hints.yaml beside NOTES.md. Keep it operational and compact:

yaml
api_imports:
  - "from vendor_api import app"
event_examples:
  entity_added:
    python: "__action = {'ok': True}"
verify_guidance:
  read: "Return non-empty rows with stable keys."
  write: "Set __action['ok'] only after the application confirms the mutation."
default_exec_timeout: 600

When any intent_signature for this vertical reaches stable status in policy://current, prompt the operator:

"<vertical>.* pipeline flywheel is stable. Consider establishing a reverse flywheel to observe operator behavior and proactively identify repetitive actions the AI can take over. If needed, invoke the operator-monitor-debug skill."

This connects the forward flywheel (AI learns to DO tasks) to the reverse flywheel (AI learns to RECOGNIZE when humans are doing those tasks repeatedly). The vertical adapter shares the same intent_signature namespace and feeds the same policy registry — a repeated operator pattern becomes pattern_pending_synthesis, Claude Code skills inspect the facts, and verified code enters the flywheel via icc_exec WAL at explore stage.

© jianzhichun, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in skills/initializing-vertical-flywheel of jianzhichun/emerge.

Open the folder on GitHubat commit 035db30

Compare with similar skills

Initializing Vertical Flywheel next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Initializing Vertical Flywheel compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Initializing Vertical Flywheel this skilljianzhichun/emerge106—~2.2kAutomated safety check: PassMIT
Web Application Testinganthropics/skills180k51 repos~966Automated safety check: PassApache-2.0
Diagnosing Bugsfossasia/eventyay-interpretation1.6k32 repos~2.1kAutomated safety check: PassApache-2.0
TDDpietheinstrengholt/rssmonster56430 repos~906Automated safety check: PassMIT
TDD WorkflowhellangleZ/burn-in-cceverywhere-ralph11211 repos~2.4kAutomated safety check: PassNone
TDDsanity-io/sanity6.4k20 repos~1kAutomated safety check: PassMIT

Similar skills

  • Web Application Testing

    anthropics/skills

    Official

    Tests local web applications with Python Playwright scripts, checking frontend behavior, capturing screenshots and reading browser console logs.

    180k GitHub starsUsed in 51 repos~966 tokens
    Testing & QAAuto-check passed
  • Diagnosing Bugs

    fossasia/eventyay-interpretation

    Diagnosis loop for hard bugs and performance regressions. An agent skill from fossasia/eventyay-interpretation.

    1.6k GitHub starsUsed in 32 repos~2.1k tokens
    Testing & QAAuto-check passed
  • TDD

    pietheinstrengholt/rssmonster

    Test-driven development. An agent skill from pietheinstrengholt/rssmonster.

    564 GitHub starsUsed in 30 repos~906 tokens
    Testing & QAAuto-check passed
  • TDD Workflow

    hellangleZ/burn-in-cceverywhere-ralph

    A skill your agent uses when writing new features, fixing bugs, or refactoring code.

    112 GitHub starsUsed in 11 repos~2.4k tokens
    Testing & QAAuto-check passed
  • TDD

    sanity-io/sanity

    Official

    Test-driven development with red-green-refactor loop. An agent skill from sanity-io/sanity.

    6.4k GitHub starsUsed in 20 repos~1k tokens
    Testing & QAAuto-check passed
  • Context Driven Development

    Ibrahim-3d/orchestrator-supaconductor

    A skill your agent uses when working with Conductor's context-driven development methodology, managing project context artifacts, or understanding the relationship between product.md, tech-stack.md…

    381 GitHub starsUsed in 9 repos~2.9k tokens
    Testing & QAAuto-check passed

More from jianzhichun/emerge

All 14 skills in this repo
  • Distilling Operator Flows

    jianzhichun/emerge

    A skill your agent uses when wiring a generic connector to capture remote operator actions and convert repeated event facts into pending pipeline work through Claude Code skills.

    106 GitHub stars~432 tokensUpdated 5 mo ago
    Auto-check passed
  • Emerge Forward Synthesis

    jianzhichun/emerge

    Distill Emerge forward flywheel WAL samples into a parameterized pipeline result.

    106 GitHub stars~585 tokensUpdated 5 mo ago
    Auto-check passed
  • Emerge Reverse Synthesis

    jianzhichun/emerge

    Distill Emerge reverse flywheel raw operator events into a structured synthesis result.

    106 GitHub stars~451 tokensUpdated 5 mo ago
    Auto-check passed
  • Operator Monitor Debug

    jianzhichun/emerge

    A skill your agent uses when the operator monitoring pipeline appears broken: EventBus has no events, PatternDetector is not firing, elicitation dialog never appears, or OperatorMonitor is silent.

    106 GitHub stars~1k tokensUpdated 5 mo ago
    Auto-check passed
  • Policy Optimization

    jianzhichun/emerge

    A skill your agent uses when policy health needs tuning: too many explore pipelines, repeated failures, stalled promotions, or noisy rollback behavior.

    106 GitHub stars~916 tokensUpdated 5 mo ago
    Auto-check passed
  • Reflection Deep Dive

    jianzhichun/emerge

    A skill your agent uses when connector flywheel history is large/noisy and lightweight reflection is insufficient.

    106 GitHub stars~325 tokensUpdated 5 mo ago
    Auto-check passed

Categories

Questions about Initializing Vertical Flywheel

What does Initializing Vertical Flywheel do?

A skill your agent uses when a user asks to initialize a domain flywheel from natural language context, especially when environment details are incomplete or mixed with execution assumptions. Initializing Vertical Flywheel is an agent skill from jianzhichun/emerge. Use when a user asks to initialize a domain flywheel from natural language context, especially when environment details are incomplete or mixed with execution assumptions.

When should I use Initializing Vertical Flywheel?

Initializing Vertical Flywheel fits situations like: A user asks to initialize a domain flywheel from natural language context; especially when environment details are incomplete; mixed with execution assumptions.

How do I install Initializing Vertical Flywheel in Claude Code?

Run `npx skills add jianzhichun/emerge --skill initializing-vertical-flywheel -a claude-code`. Or copy the skill folder (skills/initializing-vertical-flywheel in jianzhichun/emerge) into .claude/skills/initializing-vertical-flywheel in your project. Claude Code loads it when a task matches its description.

How do I install Initializing Vertical Flywheel in Codex?

Run `npx skills add jianzhichun/emerge --skill initializing-vertical-flywheel -a codex`. Or copy the skill folder (skills/initializing-vertical-flywheel in jianzhichun/emerge) into .agents/skills/initializing-vertical-flywheel in your project. Codex loads it when a task matches its description.

Can I use Initializing Vertical Flywheel in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add jianzhichun/emerge --skill initializing-vertical-flywheel -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/initializing-vertical-flywheel, .gemini/skills/initializing-vertical-flywheel, .github/skills/initializing-vertical-flywheel and .opencode/skills/initializing-vertical-flywheel in your project.

What does Initializing Vertical Flywheel need to run?

Going by SKILL.md and its folder, Initializing Vertical Flywheel needs the command-line tools its instructions call (python3 and pytest). Our summary lists: Python 3.

Does Initializing Vertical Flywheel access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Initializing Vertical Flywheel safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Initializing Vertical Flywheel use?

Initializing Vertical Flywheel is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Initializing Vertical Flywheel use?

About 2.2k tokens (SKILL.md is roughly 8.7k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Initializing Vertical Flywheel?

Skills that share tags, products or a category with Initializing Vertical Flywheel: Web Application Testing (anthropics/skills, 180k stars), Diagnosing Bugs (fossasia/eventyay-interpretation, 1.6k stars), TDD (pietheinstrengholt/rssmonster, 564 stars) and TDD Workflow (hellangleZ/burn-in-cceverywhere-ralph, 112 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Initializing Vertical Flywheel?

jianzhichun (a GitHub user) maintains it in jianzhichun/emerge, which has 106 GitHub stars. The repository holds 14 skills in this directory. The repository was last updated on April 26, 2026.

Source: jianzhichun/emerge on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.