Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/bitwikiorg/init.md/llms.txt

Use this file to discover all available pages before exploring further.

init.md is a Markdown file you copy and run — there is nothing to install, no account to create, and no build step required. You identify a target, copy the protocol or the template that best fits it, and follow the procedure. The result is a validated, operational target and a clear report of what was done.
1

Identify your target

Before copying anything, be specific about what you are initializing. Record three things:
  1. What you are initializing — give it a name and describe its purpose (e.g. “a Node.js CLI tool”, “an AI agent workspace”, “a staging server”).
  2. Where it lives — its location on disk, a repository URL, or a remote host.
  3. What “operational” means for it — a passing test suite, a running service, a complete instruction file, a validated build.
This framing is the input to the procedure. The clearer it is, the more targeted the initialization will be.
2

Choose init.md or a template

Five templates cover the most common targets. Pick the one that most closely matches your target, then adapt it as needed.
TemplateBest for
MinimalAny small target with a single clear operational gap
Dry RunInspect and plan without making any changes to the target
Development ProjectSoftware repos, apps, packages, CLIs, websites, and monorepos
AgentAI agent workspaces, agent-enabled repos, and tool-using assistant environments
ServerServer hosts, infrastructure, deployment targets, and service readiness
If no template exactly matches your target, use the root init.md itself — it defines the general procedure that all templates are built on.See the full template reference at Templates Overview.
3

Choose a mode

Modes describe how the procedure should behave. Select the mode that matches your intent before running the protocol.
ModeBehavior
ActiveInspect the target and perform all applicable initialization work.
Dry RunInspect the target and report proposed work without making any changes.
RepairInspect an existing target, identify missing or invalid elements, and correct what applies.
ReinitializeRe-evaluate an already initialized target after substantial changes to its purpose, environment, or operating model.
Most first-time initializations use Active mode. Use Dry Run when you want to preview the plan before committing to changes.
4

Copy the protocol

Visit the init.md site or this documentation to copy the content:
  • For the general method: copy the root init.md — it contains the complete target-neutral procedure.
  • For a specific target type: copy the body of the matching template — the usable instruction content, not the metadata header.
The Markdown file is the protocol. Paste it into your editor, your agent’s context, or the tool you are using to run the procedure.
5

Run the procedure

Run the procedure against your target by following the six stages yourself, or by instructing an agent to do so. The examples below show how to phrase instructions for both cases.
# For the general method:
Use the complete init.md procedure to inspect this target, initialize what applies, validate it, and report status.

# For a software repository:
Use the complete development project init template to initialize this repository and validate the developer workflow.

# For an AI agent workspace:
Use the complete agent init template only if this target needs agent instructions, tools, prompts, or state.

# For a dry-run (no changes):
Use the complete dry-run init template to inspect this target and report proposed init work without changing files.
The procedure will inspect what exists, determine what is missing, create or configure only what applies, and validate the result before reporting.
If no template exactly fits your target, copy init.md itself and write a short target-specific plan before making changes.
6

Review the validation result

Every init.md run ends with a report and one of four result states. Use these states to understand what happened and what comes next.
Result StateMeaning
OPERATIONALAll required checks passed. The target is ready.
OPERATIONAL_WITH_WARNINGSThe target is usable but one or more non-blocking issues remain.
BLOCKEDA required check failed. The target cannot be considered operational until the blocker is resolved.
DRY_RUN_COMPLETEInspection and planning finished. No changes were made to the target.
A BLOCKED result is not a failure of the procedure — it is a clear signal of exactly what needs to be resolved before the target is operational. Treat the report as the starting point for that work.

​
Next Steps

Protocol Overview

Understand the six-stage procedure in detail: what each stage inspects, creates, and validates.

Templates

Browse all five canonical templates and copy the one that matches your target.

For Humans

A step-by-step guide for running init.md yourself, choosing modes, and reviewing outputs.

For Agents

How AI agents should read, interpret, and execute init.md protocols against a target.

Build docs developers (and LLMs) love

Get started for freeTalk to us