The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from local repository files, such as READMEs and existing agent instructions, to preserve developer-authored content during the initialization process.
Ingestion points: scripts/bootstrap.py reads existing content from AGENTS.md, CLAUDE.md, and other project files via the prepare_merged function.
Boundary markers: The skill employs explicit HTML comment markers (e.g., ) to define the boundaries of managed content blocks.
Capability inventory: The skill is capable of reading from and writing to specific markdown files within the target project root, facilitating the setup of project contracts.
Sanitization: The validate_render_text function in scripts/bootstrap.py proactively validates all user-provided or extracted text to ensure it does not contain newlines or management markers, effectively preventing content from breaking out of its designated block or corrupting the configuration schema.

project-init