The Agent Skills Directory

[SAFE]: The skill performs legitimate environment configuration and guides the user to official authentication paths. It explicitly warns against using unofficial API wrappers or scripts for Linear access.
[NO_CODE]: The skill consists entirely of Markdown instructions and configuration templates; it contains no executable scripts, binaries, or logic that runs outside of the agent's instructional context.
[DATA_EXFILTRATION]: The skill accesses and modifies agent-specific configuration files (e.g., ~/.claude/CLAUDE.md, ~/.codex/AGENTS.md) solely to persist its own workflow instructions. It does not access or transmit sensitive user information, credentials, or system files.
[COMMAND_EXECUTION]: The skill utilizes platform-specific commands like cursor_dialog and open_automation to modify the agent's internal rules and schedule tasks. These interactions are confined to the agent's supported features for workspace customization.
[PROMPT_INJECTION]: The skill implements persistent "always-on" rules and response guidelines to enforce a specific communication style and workflow discipline.
Ingestion points: Instruction blocks from ALWAYS_ON.md are appended to local files like CLAUDE.md and agent rule lists via SKILL.md instructions.
Boundary markers: Content is explicitly wrapped between  markers to ensure idempotency and clear delimitation.
Capability inventory: The skill uses file-write capabilities for agent config files and the cursor_dialog tool to add user rules.
Sanitization: No automated sanitization is applied to the markdown content, as it is sourced locally from the skill package.

linear-finish-install