The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses node -e to execute several local JavaScript helper scripts (e.g., session-init-helpers.cjs, agent-files-state.cjs, project-config-loader.cjs) to assess the folder classification and verify the project state.
[COMMAND_EXECUTION]: During the documentation normalization phase, the skill executes git mv and git rm commands to reorganize files and resolve legacy naming conflicts within the repository.
[PROMPT_INJECTION]: The skill is subject to indirect prompt injection because it ingests and acts upon content from multiple untrusted project files.
Ingestion points: Processes content from docs/project-config.json, docs/project-reference/docs-index-reference.md, docs/project-reference/lessons.md, CLAUDE.md, and AGENTS.md (via SKILL.md Phase 0 and Phase 2).
Boundary markers: There are no explicit instructions or delimiters used to separate user-provided content from the agent's internal instructions when reading these files.
Capability inventory: The skill has broad capabilities including shell command execution (node, git), file system writes (git mv, git rm, updating CLAUDE.md), and the ability to spawn sub-agents (spawn_agent).
Sanitization: The skill lacks sanitization or validation of the text content ingested from these documentation files before it influences the agent's routing and planning decisions.

project-init