michael-denyer/pstack-claude / plugins/pstack/skills/poteto-mode
poteto-mode
Verifiedpoteto's agent style for concise, detailed responses, deliberate subagents, unslopped prose, simple code, and verified work. Use for poteto, /poteto-mode, or requests to work in this style.
references/grok.md
View on GitHub# Grok Build tool mapping for pstack
Grok Build loads the pstack plugin's skills. It does not translate Claude Code tool names in those skills. Use the tools the session exposes; availability and permissions can differ.
## Tool actions
| pstack / Claude action | Grok Build equivalent |
| --- | --- |
| Read (`Read`) | `read_file` with `target_file` set to the path in the session's skills list |
| Write (`Write`) / edit (`Edit`) | `write` / `search_replace` |
| Shell (`Bash`) | `run_terminal_command` |
| Search (`Grep` / `Glob`) | `grep` / `list_dir` |
| Web (`WebSearch` / `WebFetch`) | `web_search` / `run_terminal_command` with `curl` |
| Skill (`Skill`) | Read the path from the session's skills list with `read_file`; `/poteto-mode` also expands the skill |
| Subagent (`Agent` / `Task`) | `spawn_subagent` with `prompt` and `description`; `run_in_background` defaults to `true`; use `isolation: "worktree"` where needed |
| Wait for a subagent (`TaskOutput`) | `get_command_or_subagent_output` with its task ID |
| Send to a subagent (`SendMessage`) | `send_subagent_message`, off by default; `GROK_ACTIVE_AGENT_MESSAGES` or `[features] active_agent_messages` enables it. Without the tool, do not claim to have sent an update |
| Stop a subagent (`TaskStop`) | `kill_command_or_subagent` |
| Tasks (`TaskCreate` / `TaskUpdate` / `TodoWrite`) | `todo_write` |
| Questions (`AskUserQuestion`) | `ask_user_question` |
| MCP (`mcp__<server>__<tool>`) | `search_tool`, then `use_tool` |
| Wake-up (`ScheduleWakeup`) | For user-authorized recurring work, `scheduler_create` schedules a background subagent while the session is open; `scheduler_list` and `scheduler_delete` inspect or remove it. It is not a one-shot wake-up. |
| Claude transcript search | Grok stores sessions under `${GROK_HOME:-~/.grok}/sessions`; inspect the relevant session there rather than Claude's projects directory |
Grok does not inject the Claude SessionStart hook output. The Grok-only routing skill asks the model to read poteto-mode through its description; this is a model-followed hint, not guaranteed injection. The pstack `session hook: off` setting does not suppress it. Disable the plugin to remove the hint and its other skills.
## Subagent policy
poteto-mode's Subagents section dispatches `subagent_type: "pstack:poteto-agent"`, and other skills name `general-purpose`. The Grok build ships skills and no agent definitions.
- For `pstack:poteto-agent` or `general-purpose`, omit `subagent_type`. Grok cannot select a built-in type by name and runs an omitted type as `general-purpose`.
- Put the absolute path of `poteto-mode`'s `SKILL.md` in every brief that stands in for `pstack:poteto-agent`, as poteto-mode asks.
## Model names
Skills name Claude Code model aliases in their Models sections. Grok Build's available models depend on the user's configuration; do not pass a Claude alias as a Grok model.
- The model sheet is `${GROK_HOME:-~/.grok}/pstack-models.md`. Before a skill that needs a role model (`arena`, `interrogate`, `no-comments`, `poteto-mode`, `swarm`, `thermos`), read it. With no sheet, use the runtime default; run `setup-pstack` only when the user asks to configure roles.
- The `spawn_subagent` tool says to omit `model` unless the user explicitly requests one. It has no effort parameter; when it cannot apply a sheet choice, report that limitation instead of claiming model diversity.
- Roles that default to the worker model (`feature, refactoring`): the worker model the user chose.
- Roles that default to the strongest model (`strongest judgment`): the strongest model the user chose.
- Diverse-model panels (`arena`, `interrogate`) require distinct models. Report when the available dispatch cannot provide that diversity.
`/setup-pstack` writes the sheet. Choose only models the session exposes.
## Per-skill notes
| Skill | On Grok Build |
| --- | --- |
| `why` | List MCP servers with `search_tool`, not from Claude Code's MCP files. |
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |