with one click
init
// Initialize or update AGENTS.md. Guides the agent to autonomously analyze the codebase and generate a high-quality context file.
// Initialize or update AGENTS.md. Guides the agent to autonomously analyze the codebase and generate a high-quality context file.
[HINT] Download the complete skill directory including SKILL.md and all related files
| name | init |
| description | Initialize or update AGENTS.md. Guides the agent to autonomously analyze the codebase and generate a high-quality context file. |
This skill defines the standard for creating an AGENTS.md file. Instead of following a rigid script, use your judgment and exploration tools to understand the project deeply before documenting it.
Create a "readme for robots" (AGENTS.md) that accelerates understanding for future AI sessions. It must include the project's stack, structure, workflows, and coding standards.
Explore the repository to build a mental model of the project. You determine which files to read.
Create or update AGENTS.md with your findings.
Use this structure as a baseline, but adapt it if the project has unique needs.
# PROJECT KNOWLEDGE BASE
**Generated:** {CURRENT_DATE}
## OVERVIEW
Project: **{PROJECT_NAME}**
Stack: {TECHNOLOGIES_AND_VERSIONS}
## STRUCTURE
{TREE_VIEW_OR_KEY_DIRECTORIES_LIST}
* `{DIR}`: {DESCRIPTION}
## COMMANDS
| Action | Command |
|--------|---------|
| Install| `{CMD}` |
| Test | `{CMD}` |
| Build | `{CMD}` |
| Run | `{CMD}` |
## CODING STANDARDS
* **Language**: {LANGUAGE_DETAILS}
* **Style**: {OBSERVED_PATTERNS}
* **Rules**: {LINTER_OR_FORMATTER_INFO}
## WHERE TO LOOK
* **Source**: `{PATH}`
* **Tests**: `{PATH}`
* **Docs**: `{PATH}`
## NOTES
* {CRITICAL_CONTEXT_OR_GOTCHAS}
AGENTS.md exists, read it first. Merge new insights without overwriting human-authored context..cursorrules, CLAUDE.md).