The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill establishes a workflow where the agent consumes and updates docs/REQUIREMENTS.md and docs/BUSINESS-RULES.md as primary task context.
Ingestion points: Files docs/REQUIREMENTS.md and docs/BUSINESS-RULES.md are intended to be read by agents at the start of every session to obtain project requirements and domain logic.
Boundary markers: The instructions do not define delimiters or specific safety warnings to prevent the agent from obeying instructions that might be maliciously embedded in these documentation files by external contributors.
Capability inventory: The skill assumes the agent has broad capabilities, including file system access (read/write), shell command execution via bash, and the ability to interact with external issue trackers like Linear or Jira.
Sanitization: There is no process defined to sanitize or validate the content of the requirements or business rules before they are incorporated into the agent's prompt.
[COMMAND_EXECUTION]: The skill configures multiple automation points that execute shell commands and scripts.
Evidence: It sets up Git pre-commit hooks using Lefthook or Husky, which run tasks like type-checking, linting, and testing during the commit process. It also guides the agent to create and execute bash scripts for Claude Code's hook system.
[DYNAMIC_EXECUTION]: The skill provides templates for dynamic tool interception hooks and plugins.
Evidence: The skill includes templates for .claude/settings.json hooks and .opencode/plugins/ TypeScript modules. These components are designed to intercept tool calls at runtime, evaluating and potentially blocking them based on custom logic.
[EXTERNAL_DOWNLOADS]: The skill recommends the installation and use of several third-party development and source-exploration tools.
Evidence: It promotes the installation of tools such as opensrc, oxlint, oxfmt, lefthook, and biome via standard package managers. The recommended opensrc tool is used to dynamically resolve and fetch source code from registries like npm, PyPI, and crates.io, as well as directly from GitHub repositories.

ai-repo-setup