Skip to content
Back to search

oh-my-antigravity

Context-engineering-powered multi-agent team workflow pack for Antigravity CLI.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

It has changed since its scan on 11 Nov 2026.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/Joonghyun-Lee-Frieren/oh-my-antigravity with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/Joonghyun-Lee-Frieren/oh-my-antigravity'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/joonghyun-lee-frieren--oh-my-antigravity/badge.svg)](/listings/joonghyun-lee-frieren--oh-my-antigravity)

HTML

<a href="/listings/joonghyun-lee-frieren--oh-my-antigravity"><img src="/api/listings/joonghyun-lee-frieren--oh-my-antigravity/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

Requires current Antigravity CLI and Node.js 20+.

agy plugin install https://github.com/Joonghyun-Lee-Frieren/oh-my-antigravity
agy plugin list

Inside the CLI, inspect `/plugin list` and `/hooks`, then start with:

/oma-team-plan Describe the change and its acceptance criteria

The documentation reader returns a concise summary. Record a Scope Brief, seal the file scope and checks, and request an independent Design Check. After it passes, **end the design turn**. In your next turn:

/oma-team-exec Implement the sealed plan

Run `/oma-team-verify` to collect remaining section reviews. Stop schedules repair batches when checks or reviews fail. Missing services, permissions, or evidence can end in `blocked` or `unknown`; success is never a required answer.

Read the [guarded workflow guide](docs/guide/guarded-workflow.md) for exact controller commands, JSON examples, recovery, and enforcement limits.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

plugin.json · Manifest · 5 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. {
  2. "$schema": "https://antigravity.google/schemas/v1/plugin.json",
  3. "name": "oh-my-antigravity",
  4. "description": "Evidence-first engineering with Scope Brief, independent Design Check, bounded evidence, and verified session continuity."
  5. }
README.md · README · 188 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. # oh-my-antigravity (OmA)
  2. [![Version](https://img.shields.io/badge/version-1.2.0-0d8a83)](docs/history.md)
  3. [![Checks](https://img.shields.io/github/actions/workflow/status/Joonghyun-Lee-Frieren/oh-my-antigravity/version-check.yml?branch=main&label=checks)](https://github.com/Joonghyun-Lee-Frieren/oh-my-antigravity/actions)
  4. [![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](LICENSE)
  5. [Website](https://joonghyun-lee-frieren.github.io/oh-my-antigravity/) · [History](docs/history.md) · [한국어](docs/README_ko.md) · [日本語](docs/README_ja.md) · [Français](docs/README_fr.md) · [中文](docs/README_zh.md) · [Español](docs/README_es.md)
  6. A native **Antigravity CLI plugin** for engineering work with separate design turns,
  7. read-only documentation agents, gated edits, real verification evidence, and small reviews.
  8. <p align="center"><img src="resources/image/oma_logo_02.png" alt="OmA" width="360"></p>
  9. ## Install
  10. Requires current Antigravity CLI and Node.js 20+. Python 3.9+ is additionally needed
  11. when the sealed scope contains Python files.
  12. ```bash
  13. agy plugin install https://github.com/Joonghyun-Lee-Frieren/oh-my-antigravity
  14. agy plugin list
  15. ```
  16. Inside the CLI, inspect `/plugin list` and `/hooks`, then start with:
  17. ```text
  18. /oma-team-plan Describe the change and its acceptance criteria
  19. ```
  20. Use `/oma-read` to register and fully page the source snapshots, then obtain the
  21. documentation reader's concise summary. Use `/oma-probe` for required CLI help.
  22. Record a Scope Brief, seal the
  23. file scope and checks, and request an independent Design Check. After it passes,
  24. **end the design turn**. In your next turn:
  25. ```text
  26. /oma-team-exec Implement the sealed plan
  27. ```
  28. Run `/oma-team-verify` to collect remaining section reviews. Stop schedules repair batches
  29. when checks or reviews fail. Missing services, permissions, or evidence can end in
  30. `blocked` or `unknown`; success is never a required answer.
  31. Read the [guarded workflow guide](docs/guide/guarded-workflow.md) for exact controller
  32. commands, JSON examples, recovery, and enforcement limits.
  33. ## What's new in v1.2.0
  34. Validator scripts and declared check `inputs` now have recorded hashes. Changed
  35. dependencies block start, verification, or acceptance of stale evidence. Session
  36. Trail also reports validator drift. Every covering check result reaches the section
  37. reviewer; up to two bounded log excerpts prioritize failures. See the
  38. [input schema and limits](docs/guide/guarded-workflow.md) for legitimate scoped test
  39. changes and dependency declarations. Existing sealed plans need resealing for the
  40. new input protection.
  41. The supplied Part 1–7 architecture now also drives enforced source-read coverage,
  42. read-only dependency fingerprints, large-file edit limits, typed failure exits,
  43. Python AST validation, dependency-ordered repair groups, and one-group focus on
  44. repair oscillation. Markdown reviews split at headings, use a fresh conversation
  45. per shard, and require disconfirmation evidence before a deterministic judge decides.
  46. Start with `/oma-read`, `/oma-probe`, `/oma-failure`, and `/oma-repair-context`.
  47. Read the [Part 1–7 mapping and complete schemas](docs/guide/series-runtime.md),
  48. including migration and enforcement limits. These safeguards do not guarantee
  49. zero defects or prove that a model understood the delivered source.
  50. ## Scope and continuity in v1.1.0
  51. | Surface | Behavior |
  52. | --- | --- |
  53. | **Scope Brief** | Record the objective, constraints, non-goals, decision owners, and source conflicts; unresolved items block sealing |
  54. | **Design Check** | A read-only auditor checks the exact plan before the design turn can close; plan changes invalidate prior review |
  55. | **Evidence Window** | Read integrity-checked log excerpts with a bounded UTF-8 byte cursor; full output remains on disk |
  56. | **Session Trail** | Immutable JSON/Markdown checkpoints retain the current intent and next action; resume detects source/plan drift |
  57. | **Native review receipts** | The reviewer submits its own result from another conversation; the parent cannot author PASS or steer the review |
  58. | **Care profiles** | Routine/elevated/critical scopes use 120/90/60-line sections and 3/2/1-unit batches, while keeping all checks and locks |
  59. Sensitive paths can only raise care. Compact status/context packs keep full logs and
  60. receipt tickets out of routine model context. Checks still run and missing proof
  61. stays unknown. Read [schemas and examples](docs/guide/session-continuity.md).
  62. ## Models
  63. Recommendations come from [config/models.json](config/models.json), checked against the [official model documentation](https://antigravity.google/docs/models) on 2026-11-16.
  64. | Lane | Carbon | Balanced | Hydrogen | Flash |
  65. | --- | --- | --- | --- | --- |
  66. | Design/planning | `gemini-4-carbon-high` | `gemini-4-argon-high` | `gemini-4-hydrogen-high` | `gemini-3.8-flash-high` |
  67. | Implementation | `gemini-4-hydrogen-medium` | `gemini-4-hydrogen-medium` | `gemini-4-hydrogen-medium` | `gemini-3.8-flash-medium` |
  68. | Quick work | `gemini-4-hydrogen-low` | `gemini-4-hydrogen-low` | `gemini-4-hydrogen-low` | `gemini-3.8-flash-low` |
  69. | Review/verification | `gemini-4-carbon-high` | `gemini-4-argon-high` | `gemini-4-hydrogen-high` | `gemini-3.8-flash-high` |
  70. Use `agy models` to check account availability, then select a model with `agy --model <id>` or `/model`. For example, `agy --model gemini-4-hydrogen-xhigh` selects Hydrogen's extra-high reasoning level. Carbon is available on Ultra and eligible Enterprise accounts; Argon availability depends on the plan. Choose an available profile explicitly if necessary.
  71. `/oma-model` explains the profiles. `auto` keeps the selected runtime model; `custom` requires all four lanes. These are recommendations, not automatic model switches: hooks cannot override the runtime model and subagents use `model: inherit`. Antigravity's native [Orchestrator](https://antigravity.google/docs/models#orchestrator) is a separate runtime option that coordinates Carbon or Argon with Hydrogen and has a fixed effort setting.
  72. ## Workflows and roles
  73. The plugin exposes 64 short skills, including 53 `/oma-<name>` command entrypoints,
  74. and 16 capability-limited agents. Use only the procedures relevant to the current task.
  75. | Command | Purpose |
  76. | --- | --- |
  77. | `/oma-read`, `/oma-probe` | Full source-window coverage and actual bounded CLI introspection |
  78. | `/oma-failure`, `/oma-repair-context` | Typed honest failure and active dependency-ordered repair diagnostics |
  79. | `/oma-team-plan`, `/oma-team-prd` | Documentation handoff, scope, acceptance criteria, and design |
  80. | `/oma-brief` | Objective, constraints, decision ownership, and source conflicts |
  81. | `/oma-design-check` | Independent check of the exact sealed plan |
  82. | `/oma-evidence` | Bounded, integrity-checked validator log windows |
  83. | `/oma-checkpoint`, `/oma-resume` | Immutable handoff and read-only continuity inspection |
  84. | `/oma-team-exec` | Apply the sealed scope in the next user turn |
  85. | `/oma-team-verify` | Real checks and one-section review handoffs |
  86. | `/oma-team-fix` | The Stop-selected batch of up to three repair units |
  87. | `/oma-goal`, `/oma-ultragoal` | Delivery tracking with the same turn and evidence gates |
  88. | `/oma-status`, `/oma-doctor` | Phase, ownership, evidence gaps, locks, and blockers |
  89. | `/oma-model` | Current model recommendations and runtime selection |
  90. | `/oma-hooks`, `/oma-hooks-validate` | Native hook inspection and safety diagnostics |
  91. | `/oma-stop`, `/oma-cancel` | Stop with evidence and locks retained |
  92. `oma-doc-reader` reads specifications. `oma-planner` and `oma-architect` help design.
  93. `oma-executor` proposes patches; only the owning main conversation applies them.
  94. `oma-design-auditor` checks the sealed plan. `oma-reviewer` examines one bounded section; `oma-verifier` checks acceptance evidence.
  95. Every shipped worker excludes write, shell, MCP, and delegation tools.
  96. The authoritative workflow state is `.omg/state/guard.json`; execution logs live in
  97. `.omg/evidence/`. Agents use `.omg/drafts/` for requests and auxiliary notes. The controller
  98. alone owns guard state, evidence, and checkpoints. Independent reviewer messages are
  99. captured by the hook using the real sender identity. One workspace and one writer are allowed per workflow;
  100. use separate workflows for isolated worktrees.
  101. ## Upgrade from v1.0 or Gemini CLI
  102. Reinstall the native plugin using `agy plugin install`. New plans require a settled
  103. Scope Brief and native Design Check. Independent roles need native send_message
  104. hooks with the actual child conversation ID; missing runtime support stays unknown.
  105. Existing v1.0 sealed state can finish through its legacy receipt path, but new plans
  106. always use the native path. Session Trail inspects current state without restoring
  107. old ownership, approval, or locks.
  108. Native commands use
  109. `/oma-team-exec`, replacing `/oma:team-exec`. The old TOML protocol sources live in
  110. `protocols/oma/` and generate native skill wrappers; they are outside the runtime's
  111. legacy `commands/` discovery directory to avoid duplicate registrations.
  112. `gemini-extension.json` remains a migration marker. Remove manually copied old hooks.
  113. The Gemini `BeforeModel` router, router trace flags, and `AfterAgent` learn nudge are
  114. removed. Use `/oma-learn` explicitly and select models through the CLI.
  115. Passed files remain locked, including across subsequent workflows. Reopening a lock,
  116. changing sealed scope, or recovering blocked/foreign-owned state requires an explicit
  117. operator action outside the guarded agent. Preserve the prior evidence when recovering.
  118. The harness constrains agent tools; it is not an OS sandbox. Validators are trusted
  119. project code. Behavioral checks and reviewers still need to catch semantic fabrication.
  120. A repair unit is one failed validator or one non-passing section, which may contain
  121. several diagnostics. See the [full contract](docs/guide/guarded-workflow.md).
  122. ## Development
  123. ```bash
  124. npm test
  125. npm run test:skills
  126. npm run test:generated
  127. npm run test:version
  128. agy plugin validate .
  129. ```
  130. Edit command procedures in `protocols/oma/`, then run `npm run build:skills`.
  131. Core skill procedures live beside their entrypoints in `references/`.
  132. Tests cover turn boundaries, shell/path bypasses, ownership, failed/missing checks,
  133. review coverage, stale evidence, locks, repair limits, cancellation, and retry bounds.
  134. v1.1 also covers source conflicts, design receipt invalidation, actor spoofing, replay,
  135. care escalation, UTF-8 log paging, checkpoint corruption, and source drift on resume.
  136. ## Special thanks
  137. **Special thanks to [@hyunuk.lim](https://www.threads.com/@hyunuk.lim) (threads).** His threads and posts were a major
  138. help in shaping this refactor: smaller skills, physical tool constraints, honest
  139. failure, bounded repairs, and focused reviews.
  140. Further reading: [his Medium article on the ideas behind the v1.0 renewal](https://medium.com/@dev.hyunuk/3f86bb9b8ab7).
  141. Thank you for sharing the concepts that helped shape OmA's engineering workflow.
  142. The project also grew from Jeongkyu Shin's observation that the engineering harness
  143. around a model matters alongside the model itself.
  144. [MIT License](LICENSE) · [Sponsor](https://github.com/sponsors/Joonghyun-Lee-Frieren)

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.