Init Project
Agent Buildinginit-projectInit Project is an Agent Skill for agent building published in the UfoMiao/zcf repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.
Initialize project AI context, generate/update root-level and module-level CLAUDE.md indexes
QUICK START
How to use this skill
Bring this guide into your coding agent with a prompt tailored to the tool you use.
- Open your project in Codex.
- Copy and paste the prompt below.
- Review files and risks before approving any action.
Prompt to paste
Inspect the Agent Skill at https://github.com/UfoMiao/zcf/blob/HEAD/templates/skills/en/init-project/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/init-project, but wait for my approval before writing files or running scripts.
Copying this prompt does not install or run the skill. Review third-party files before use.
Usage
/init-project <PROJECT_SUMMARY_OR_NAME>
Objective
Initialize project AI context using a mixed strategy of "concise at root + detailed at module level":
- Generate/update
CLAUDE.mdat repository root (high-level vision, architecture overview, module index, global standards). - Generate/update local
CLAUDE.mdin identified module directories (interfaces, dependencies, entry points, tests, key files, etc.). - ✨ For improved readability, automatically generate Mermaid structure diagrams in the root
CLAUDE.mdand add navigation breadcrumbs to each moduleCLAUDE.md.
Orchestration Instructions
Step 1: Call the get-current-datetime sub-agent to obtain the current timestamp.
Step 2: Call the init-architect sub-agent once, with input:
project_summary: $ARGUMENTScurrent_timestamp: (timestamp from step 1)
Execution Strategy (Agent adapts automatically, no user parameters needed)
- Stage A: Repository Census (Lightweight) Quickly count files and directories, identify module roots (package.json, pyproject.toml, go.mod, apps/, packages/, services/*, etc.).
- Stage B: Module Priority Scanning (Medium) For each module, perform targeted reading and sampling of "entry/interfaces/dependencies/tests/data models/quality tools".
- Stage C: Deep Supplementation (As Needed) If repository is small or module scale is small, expand reading scope; if large, perform batch supplemental scanning on high-risk/high-value paths.
- Coverage Measurement and Resumability
Output "scanned files / estimated total files, covered module ratio, ignored/skipped reasons" and list "recommended next-step deep-dive sub-paths". When running
/init-projectrepeatedly, perform incremental updates and breakpoint resumable scanning based on previous index.
Security and Boundaries
- Only read/write documentation and indexes, do not modify source code.
- Ignore common generated artifacts and binary large files by default.
- Print "summary" in main dialog, write full content to repository.
Output Requirements
- Print "Initialization Result Summary" in main dialog, including:
- Whether root-level
CLAUDE.mdwas created/updated, major section overview. - Number of identified modules and their path list.
- Generation/update status of each module's
CLAUDE.md. - ✨ Explicitly mention "Generated Mermaid structure diagram" and "Added navigation breadcrumbs for N modules".
- Coverage and major gaps.
- If not fully read: explain "why stopped here" and list recommended next steps (e.g., "suggest priority supplemental scanning: packages/auth/src/controllers").
- Whether root-level