drn

Safely add a directory to Claude Code's Bash tool PATH. Use when a binary runs in your terminal but is "command not found" inside Claude Code's Bash tool, when adding a directory to PATH for Claude Code, when configuring CLAUDE_ENV_FILE, or when settings.json env.PATH did not work. Explains why env.PATH replaces rather than appends PATH and is not variable-expanded, why ~/.zshrc PATH exports are dropped, and the supported SessionStart hook fix.

Create or improve a Claude Code skill/slash command with best practices for structure, dynamic context, and safety. Use for creating new skills, improving existing ones, or learning skill authoring.

Launch a dynamic Workflow where the top-tier session model (Fable, or Opus when Fable is unavailable) handles planning and orchestration while implementation subagents run on Sonnet for routine tasks and Opus for complex ones. Use when the user wants to orchestrate a build, a dynamic workflow, a model-tiered build, fable or opus planning with sonnet and opus implementation, or tiered agents.

Open a PR, wait for CI to pass, fix failures, address review comments, and loop until fully green. Use when opening a PR, fixing CI, or addressing review feedback.

Manage recurring tasks in the local Argus daemon via its HTTP API. Use when the user wants to schedule a task locally in argus, create a cron-driven task, fire X every weekday/hour/morning, set up a recurring agent that needs local filesystem access (logs in ~/.argus, local databases, dotfiles), list or update existing argus schedules, or run an argus schedule now. Distinct from /schedule, which creates remote cloud routines without local access.

Generate a handoff prompt to pass context to another agent thread. Use when switching repos, handing off work, or sharing context between agents.

Merge current branch to the default branch via GitHub PR merge. Use when ready to merge a PR or land a branch.

Generate SVG logo alternatives with side-by-side comparison. Use when creating logos, branding, or project icons.

The hera plan-DAG: author staged, dependency-ordered multi-worker plans inside an argus sandbox and let the daemon gater materialize each node into a born-bound worker as its blockers finish. Load this (in addition to the base `hera` skill) when you are a coordinator and the work decomposes into multiple hera worker units that have DEPENDENCIES among them (one stage needs another's output, or a required ordering) — internal dependencies are the clean signal to plan a DAG rather than spawn workers ad hoc. Covers the authoring/mutation verbs, gating-on-`done`, automatic branch-stacking and its fan-in footgun, short-id node naming, sub-coordinator nodes, and the self-guard prompt patterns. NOT for non-argus sessions; NOT for in-session ephemeral work (use Claude's native sub-agents); NOT for independent workers with no ordering (just spawn them).

Inside an argus sandbox (cwd under ~/.argus/worktrees/ or ARGUS_TASK_ID set), coordinate multi-agent work via hera's mcp__argus__hera_* tools — bootstrap an orchestrator, claim or attach a worker/freelance role, spawn workers, and message roles over the idle-gated bus. Load this skill when you are (or are becoming) a coordinator, when you're spawned as a hera worker, or before delegating work that must run as its OWN argus session (separate worktree / long-running agent / own PR / own sandbox) — as opposed to in-process Claude sub-agents, which stay the right tool for ephemeral in-session fan-out. For staged or dependency-ordered multi-worker plans, this skill routes you to the companion `hera-plan` skill. NOT for non-argus sessions, where these MCP tools are not registered.

The hera plan-DAG: author staged, dependency-ordered multi-worker plans inside an argus sandbox and let the daemon gater materialize each node into a born-bound worker as its blockers finish. Load this (in addition to the base `hera` skill) when you are a coordinator and the work decomposes into multiple hera worker units that have DEPENDENCIES among them (one stage needs another's output, or a required ordering) — internal dependencies are the clean signal to plan a DAG rather than spawn workers ad hoc. Covers the authoring/mutation verbs, gating-on-`done`, automatic branch-stacking and its fan-in footgun, short-id node naming, sub-coordinator nodes, and the self-guard prompt patterns. NOT for non-argus sessions; NOT for in-session ephemeral work (use Claude's native sub-agents); NOT for independent workers with no ordering (just spawn them).

Inside an argus sandbox (cwd under ~/.argus/worktrees/ or ARGUS_TASK_ID set), coordinate multi-agent work via hera's mcp__argus__hera_* tools — bootstrap an orchestrator, claim or attach a worker/freelance role, spawn workers, and message roles over the idle-gated bus. Load this skill when you are (or are becoming) a coordinator, when you're spawned as a hera worker, or before delegating work that must run as its OWN argus session (separate worktree / long-running agent / own PR / own sandbox) — as opposed to in-process Claude sub-agents, which stay the right tool for ephemeral in-session fan-out. For staged or dependency-ordered multi-worker plans, this skill routes you to the companion `hera-plan` skill. NOT for non-argus sessions, where these MCP tools are not registered.

Manage recurring or one-shot tasks in the local Argus daemon via its HTTP API. Use when the user wants to schedule a task locally in argus, create a cron-driven task, fire X every weekday/hour/morning, fire something once at a specific future time (one-shot via run_once_at), set up an agent that needs local filesystem access (logs in ~/.argus, local databases, dotfiles), list or update existing argus schedules, or run an argus schedule now. Distinct from /schedule, which creates remote cloud routines without local access.

Mark the current Argus task as complete. Use when the work for the current worktree is done and the user wants the task to transition to the "complete" status.

Plan and execute a new feature phase for Anthesis using the proven pinned-contract parallel workflow pattern. Use when starting a new major feature or phase.

Run Anthesis windowed with a SceneTree harness script to verify a change live: drive gameplay via command bus, capture screenshots, check for script errors. Use when asked to verify a change works in the real game, take screenshots, or test gameplay end-to-end.