Skip to content

michael-denyer/pstack-claude / plugins/pstack/skills/poteto-mode

poteto-mode

Verified

poteto'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.

FileKindSize
SKILL.mdskill md18.2 KB
playbooks/authoring-a-skill.mdmarkdown852 B
playbooks/autonomous-run.mdmarkdown1.9 KB
playbooks/autopilot-full.mdmarkdown8.3 KB
playbooks/autopilot-stack.mdmarkdown1.3 KB
playbooks/babysit.mdmarkdown2.3 KB
playbooks/bug-fix.mdmarkdown4.8 KB
playbooks/eval.mdmarkdown2.8 KB
playbooks/feature.mdmarkdown3.4 KB
playbooks/hillclimb.mdmarkdown4.4 KB
playbooks/investigation.mdmarkdown920 B
playbooks/multi-phase-plan.mdmarkdown1.6 KB
playbooks/opening-a-pr.mdmarkdown1.0 KB
playbooks/pause-safely.mdmarkdown1.9 KB
playbooks/perf-issue.mdmarkdown1.5 KB
playbooks/prototype.mdmarkdown2.1 KB
playbooks/refactoring.mdmarkdown3.3 KB
playbooks/runtime-forensics.mdmarkdown1.3 KB
playbooks/session-pickup.mdmarkdown2.7 KB
playbooks/shipping.mdmarkdown7.2 KB
playbooks/trace-forensics.mdmarkdown1.8 KB
playbooks/visual-parity.mdmarkdown1.2 KB
playbooks/worktree-cleanup.mdmarkdown4.5 KB
references/agents/comment-sicko.mdmarkdown2.9 KB
references/agy-tools.mdmarkdown10.3 KB
references/bugbot-triage.mdmarkdown9.4 KB
references/codex-tools.mdmarkdown13.6 KB
references/copilot-tools.mdmarkdown20.6 KB
references/dsh-tools.mdmarkdown4.8 KB
references/grok.mdmarkdown3.9 KB
references/kimi-tools.mdmarkdown4.6 KB
references/licenses/LICENSEother1.1 KB
references/licenses/LICENSE-cursor-team-kitother1.0 KB
references/licenses/LICENSE-thermosother1.0 KB
references/licenses/NOTICE.mdmarkdown1.9 KB
references/merge-safety.mdmarkdown5.2 KB
references/minimax-tools.mdmarkdown10.2 KB
references/muse-tools.mdmarkdown12.5 KB
references/omp-tools.mdmarkdown6.8 KB
references/opencode-tools.mdmarkdown3.3 KB
references/pi-tools.mdmarkdown23.5 KB
references/resume-storage.mdmarkdown3.6 KB
scripts/.prettierignoreother652 B
scripts/.prettierrc.jsonconfig29 B
scripts/bootstrap.tsscript2.0 KB
scripts/bun.lockother6.5 KB
scripts/check-playbooks.mjsscript6.8 KB
scripts/codex-session-effort.mjsscript3.9 KB
scripts/package.jsonmanifest320 B
scripts/resume.mjsscript7.2 KB
scripts/tsconfig.jsonconfig304 B
scripts/watch-pr/check-suites.tsscript1.9 KB
scripts/watch-pr/cli.port.test.tsscript2.9 KB
scripts/watch-pr/cli.test.tsscript11.2 KB
scripts/watch-pr/cli.tsscript8.7 KB
scripts/watch-pr/deadline.tsscript522 B
scripts/watch-pr/fakes.test-helper.tsscript4.2 KB
scripts/watch-pr/fresh-install.test.tsscript1.3 KB
scripts/watch-pr/github.port.test.tsscript19.7 KB
scripts/watch-pr/github.test.tsscript16.6 KB
scripts/watch-pr/github.tsscript34.7 KB
scripts/watch-pr/head-checks.tsscript1.3 KB
scripts/watch-pr/integrity.test.tsscript24.3 KB
scripts/watch-pr/landing.tsscript3.0 KB
scripts/watch-pr/live-merge-safety.mjsscript7.4 KB
scripts/watch-pr/policy.port.test.tsscript29.0 KB
scripts/watch-pr/policy.test.tsscript14.8 KB
scripts/watch-pr/policy.tsscript29.6 KB
scripts/watch-pr/render.tsscript9.6 KB
scripts/watch-pr/ship-prexecutablescript220 B
scripts/watch-pr/shipping-cli.test.tsscript5.3 KB
scripts/watch-pr/shipping-cli.tsscript2.3 KB
scripts/watch-pr/shipping.test-helper.tsscript2.1 KB
scripts/watch-pr/shipping.test.tsscript11.0 KB
scripts/watch-pr/shipping.tsscript6.7 KB
scripts/watch-pr/transport.test.tsscript15.2 KB
scripts/watch-pr/tsconfig.jsonconfig270 B
scripts/watch-pr/types.compile.tsscript2.7 KB
scripts/watch-pr/types.tsscript15.0 KB
scripts/watch-pr/watch-prexecutablescript211 B
scripts/watch-pr/wrapper.test.tsscript587 B
scripts/worktree-audit.mjsexecutablescript22.4 KB
worker.mdmarkdown1.3 KB

references/agy-tools.md

View on GitHub
# Antigravity tool mapping for pstack

The Antigravity CLI (`agy`) loads the native pstack plugin manifest, the skills, the four plugin subagents, and the `SessionStart` hook. Resolve each Claude Code tool, model, and path a pstack skill names here.

## Tool actions

| pstack / Claude action | Antigravity equivalent |
|------------------------|------------------------|
| Read a file | `read_file` |
| Create a file | `write_file` |
| Edit a file | `edit_file` |
| Run a shell command | `run_command` (and `manage_task` to inspect, send input to, or kill background tasks) |
| Search file contents / find files | Run `rg` to search file contents and `rg --files` or `fd` to find files via `run_command` |
| Fetch a URL / search the web | `search_web` or `curl` via `run_command` |
| Invoke a skill (the `Skill` tool, `/command`) | Use `read_file` on the path in the `<skills>` catalog. Read `poteto-mode/SKILL.md` in full: the native loader includes only the first 24,000 bytes of a skill body. |
| Dispatch a subagent (the `Agent`/`Task` tool) | `invoke_subagent` with `TypeName`, `Role`, `Prompt`, `Model`, and `Workspace` |
| Dispatch N parallel subagents in one turn | N `invoke_subagent` calls in one response |
| Wait for a subagent result | End the turn (call no more tools). Antigravity resumes the session automatically when a subagent sends a message or finishes. |
| Continue or steer a subagent (`SendMessage`) | `send_message` with `Recipient` set to the subagent's `conversationId` and `Message` |
| List or stop subagents | `manage_subagents` with `Action: "list"`, `"kill"`, or `"kill_all"` and `ConversationIds` |
| Track tasks (the todolist; `TaskCreate` / `TaskUpdate`, or `TodoWrite` on Claude Code) | Keep an uncommitted `todo.md` Markdown checklist in the work dir (or `.audit/<task-slug>.todo.md` when sessions share a checkout) |
| Ask the human a fixed-choice question (`AskUserQuestion`) | `ask_question` with `questions` (`question`, `options`, `is_multi_select`) |
| Schedule a self-paced re-invocation (`ScheduleWakeup`, the `loop` skill) | `schedule` with `DurationSeconds` (one-shot timer) or `CronExpression` (recurring schedule) and `Prompt` |
| Call an MCP tool (`mcp__<server>__<tool>`) | Call the MCP tool directly as declared in the session's tool list |
| Search Claude Code transcripts under `${CLAUDE_CONFIG_DIR:-$HOME/.claude}/projects/` | Read Antigravity transcripts under `~/.gemini/antigravity-cli/brain/<conversation-id>/.system_generated/logs/transcript.jsonl` (or `antigravity` / `antigravity-ide` under `~/.gemini/`). Do not read Claude Code's projects directory. |

## Subagent policy

poteto-mode's Subagents section sets Claude-specific defaults (`subagent_type: "pstack:poteto-agent"`, `run_in_background: true`). On Antigravity:

- The plugin's four agent files (`pstack:poteto-agent` or `poteto-agent`, `pstack:comment-sicko` or `comment-sicko`, `pstack:thermo-nuclear-review-subagent` or `thermo-nuclear-review-subagent`, and `pstack:thermo-nuclear-code-quality-review-subagent` or `thermo-nuclear-code-quality-review-subagent`) register in `<subagents>` without a `tools:` frontmatter list, so Antigravity loads them as read-only without `write_file`, `edit_file`, `run_command`, or MCP tools.
- `pstack:poteto-agent` and `general-purpose` map to `TypeName: "self"` when the subagent writes files, runs commands, or uses MCP tools. Use `TypeName: "poteto-agent"` or `TypeName: "research"` only for read-only work that needs no shell or MCP access.
- Give each `poteto-mode` delegate the installed path to `poteto-mode/SKILL.md` from the `<skills>` catalog. Expand `~` to the user's home directory and tell the subagent in `Prompt` to read the entire file with `read_file`, including the model and reasoning effort sections at the end.
- `invoke_subagent` always runs asynchronously in the background, so `run_in_background: true` needs no extra flag.
- `readonly: true` (reviewers, explorers, critics) maps to `TypeName: "research"` or one of the read-only plugin agent types when no shell, MCP, or write tools are needed. When a read-only subagent needs `run_command` or MCP tools, use `TypeName: "self"` and tell it in `Prompt` not to write files.
- `isolation: "worktree"` maps to `Workspace: "share"` (creates a shared git worktree-style workspace) or `Workspace: "branch"` on `invoke_subagent`, or a manual `git worktree add` path named in `Prompt`.
- Keep the rest of the policy unchanged. Pass file pointers not inlined context, and review every subagent's diff yourself.

## Model names

Skills name Claude Code model aliases in their Models sections. On Antigravity (`agy`), `invoke_subagent`'s `Model` parameter accepts `inherit` (default, uses the parent's model), `flash_lite`, `flash`, and `pro`.

- The model sheet is `~/.gemini/config/pstack-models.md`. Before a skill that needs a role model (`arena`, `interrogate`, `no-comments`, `poteto-mode`, `swarm`, `thermos`), read it. With no sheet, read `setup-pstack`'s `SKILL.md` and finish it first.
- Pass the role's model as `Model` on `invoke_subagent`. For `inherit-parent` or `auto`, omit `Model`.
- 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`): the adversarial signal comes from model diversity. Use distinct available models and report when the panel shares a model family.

`/setup-pstack` writes the sheet. Antigravity accepts `flash_lite`; the Copilot sheet checker rejects underscores in model IDs and does not apply to Antigravity.

## Session routing hook

Antigravity runs the plugin's `SessionStart` hook when a new session starts. The hook outputs an `injectSteps` `systemMessage` JSON payload containing the Antigravity routing mandate unless `~/.gemini/config/pstack-models.md` contains `session hook: off`.

## Driver and bundled skills pstack references

The [driver policy](../SKILL.md#non-negotiables) selects the app driver. For skills and drivers named by these workflows, use these Antigravity equivalents:

| Skill or driver named in pstack | On Antigravity |
|---------------------------------|----------------|
| `run` (drive a CLI/TUI to see a change work) | Run the app yourself via `run_command` and observe the real output. |
| Project UI driver | Use the project's verification skill under `.agents/skills/` or `.claude/skills/`, or drive the UI with browser/CLI tools. |
| `plugin-dev:skill-development` (Claude's SKILL.md authoring guidance) | Read Antigravity's built-in `agy-customizations` skill (`docs/skills.md`). Keep `name` plus `description` frontmatter and progressive disclosure under 24 KB per `SKILL.md`. |
| `loop` (recurring/self-paced re-invocation, used by `babysit`) | Use the `schedule` tool with `DurationSeconds` or `CronExpression`. |

Project skills, the ones `create-verification-skill` writes under `.claude/skills/`, load from `.agents/skills/` (or `.claude/skills/` when registered) in the workspace.

## Per-skill notes

Most skills need only the tables above. These need one more mapping:

| Skill | On Antigravity |
|-------|----------------|
| `interrogate` | Each adversary is an `invoke_subagent` call with `Model` from the sheet. |
| `setup-pstack` | The [Other runtimes](../../setup-pstack/references/runtimes.md) table gives the sheet path. Use the `Model` values listed in [Model names](#model-names). |
| `architect` | A large or contested design goes through the **arena** skill: one `invoke_subagent` call per runner, each with a `Model` from the sheet's `arena runners` line. |
| `how` | The one read-only subagent maps to `invoke_subagent`. |
| `arena` | The parallel candidates and the cross-judge map to `invoke_subagent` calls with the sheet's panel lines. |
| `no-comments` | Dispatch `comment-sicko` via `invoke_subagent` with `TypeName: "comment-sicko"`. |
| `thermos` | Dispatch both review subagents via `invoke_subagent` with `TypeName: "thermo-nuclear-review-subagent"` and `"thermo-nuclear-code-quality-review-subagent"`, with the full diff and file excerpts in `Prompt`, since both types are read-only. |
| `swarm` | Each worker is an `invoke_subagent` call; use `Workspace: "share"` or `git worktree add` for writing workers. |
| `teach` | Running `how` and `why` in parallel maps to two `invoke_subagent` calls in one response; image generation uses `image-generator` via `invoke_subagent`. |
| `why` | List MCP servers from the tools Antigravity exposes to the session, not from Claude Code's MCP files. |
| `recall` | Read Antigravity transcripts under `~/.gemini/antigravity-cli/brain/<conversation-id>/.system_generated/logs/transcript.jsonl` instead of `${CLAUDE_CONFIG_DIR:-$HOME/.claude}/projects/`. |
| `reflect` | Pass the current session's transcript path (`<appDataDir>/brain/<conversation-id>/.system_generated/logs/transcript.jsonl`) to the transcript reading the skill does. |
| `show-me-your-work` | Check the log against this session's transcript or context, not a Claude Code transcript. |
| `automate-me` | Follow the `agy-customizations` skill alongside `unslop`, and write the mode skill under `.agents/skills/` or `~/.gemini/config/skills/`. |
| `create-verification-skill` | Write the generated skill where Antigravity discovers project skills, `.agents/skills/verify/`, instead of `.claude/skills/verify/`. |
| `maintain-verification-skill` | The parallel per-feature source readers map to `invoke_subagent` fan-out; the project-local skill lives under `.agents/skills/`. |
| `create-formal-verification-skill` | Write the generated skill to `.agents/skills/<area>/`, named for the area it verifies. |
| `maintain-formal-verification-skill` | Project skills, the formal verification skill among them, live under `.agents/skills/`. |
| `babysit` | `loop` and `AskUserQuestion` resolve through the tables above (`schedule` and `ask_question`). |

## Vendored scripts

Use `run_command` to run the scripts under `skills/poteto-mode/scripts/`. The `watch-pr/watch-pr` watcher needs Bun and the GitHub CLI (`gh`). Run `worktree-audit.mjs` with Node.js. It uses `gh` to look up each branch's PR and imports the transcript walker from the sibling `reflect` skill. It does not read Antigravity transcripts.

## Instructions file

Where a pstack skill says "your instructions file", on Antigravity that is `AGENTS.md` or `GEMINI.md` in the workspace, or `~/.gemini/config/AGENTS.md` globally. On Claude Code it is `CLAUDE.md`.
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: 471kUpdated Claude CodeCodexVerifiedInstall
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: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
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 availableStars: 322kUpdated Claude CodeCodexVerifiedInstall