docs: add Antigravity CLI references to documentation files

Match the same documentation coverage as the Copilot harness PR:
- ARCHITECTURE.md: add antigravity.py to adapters list, .antigravity/ to
  generated artifacts, adapter table row, update harness count
- CONTRIBUTING.md: add Antigravity CLI to harness list and portability
  checklist
- README.md: add badge, update harness list, table, Pensyve integration
- docs/authoring.md: add Antigravity column to model alias table, update
  harness count and slash-commands note
- docs/round-trip-results.md: add summary row, reproduce recipe, update
  validation command and coverage limits

Author: web-flow

ARCHITECTURE.md

@@ -4,7 +4,7 @@ Top-level architectural map for the claude-agents marketplace. Detail lives in [
 
 ## Invariants
 
-1. **Single source of truth.** All agent / skill / command authoring happens under `plugins/<name>/`. Generated harness-specific artifacts (`.codex/`, `.cursor-plugin/`, `.opencode/`, `.copilot/`, `commands/`, `agents/`, `skills/` at extension root for Gemini) are produced by adapters and gitignored. Never hand-edit generated files.
+1. **Single source of truth.** All agent / skill / command authoring happens under `plugins/<name>/`. Generated harness-specific artifacts (`.codex/`, `.cursor-plugin/`, `.opencode/`, `.copilot/`, `.antigravity/`, `commands/`, `agents/`, `skills/` at extension root for Gemini) are produced by adapters and gitignored. Never hand-edit generated files.
 
 2. **One canonical context file.** `AGENTS.md` at repo root is the only context file authored directly. `CLAUDE.md` imports it via `@AGENTS.md`. Gemini CLI reads it via `.gemini/settings.json` `context.fileName`. Codex / Cursor / OpenCode read `AGENTS.md` natively.
 
@@ -36,7 +36,7 @@ claude-agents/
 │   ├── adapters/                   # Per-harness adapter framework
 │   │   ├── base.py                 # Parser, HarnessAdapter ABC, helpers
 │   │   ├── capabilities.py         # Capability matrix; consumed by every adapter
-│   │   ├── codex.py / cursor.py / opencode.py / gemini.py / copilot.py
+│   │   ├── codex.py / cursor.py / opencode.py / gemini.py / copilot.py / antigravity.py
 │   │   └── cursor_rules/           # Hand-curated .mdc rules
 │   ├── generate.py                 # Unified CLI: `make generate HARNESS=<x>`
 │   ├── validate_generated.py       # Structural validation
@@ -62,6 +62,7 @@ Each adapter consumes the canonical `plugins/` source and emits harness-native a
 | `cursor.py` | `.cursor-plugin/`, `.cursor/rules/*.mdc` | Marketplace manifests + hand-curated rules. Cursor reads `.claude/` directly for skills/agents |
 | `opencode.py` | `.opencode/agents/`, `.opencode/commands/`, `.opencode/skills/` | Permission block from `tools:` allowlist (locked agents preserve intent); strict lowercase tool names; OpenCode-safe skill names |
 | `copilot.py` | `.copilot/agents/`, `.copilot/skills/`, `.copilot/commands/` | Markdown agent profiles + SKILL.md skills + commands-as-skills; model maps to GPT-5 family |
+| `antigravity.py` | `.antigravity/agents/`, `.antigravity/skills/`, `.antigravity/commands/` | Lowercase agent.json per agent + SKILL.md skills + commands-as-skills; model maps to Gemini 2.5 family |
 | `gemini.py` | `skills/`, `agents/`, `commands/*.toml` at extension root | Native skills + April-2026 subagents; `@{path}` injection for large command bodies |
 
 Detail in [`docs/harnesses.md`](docs/harnesses.md) (capability matrix per harness) and [`docs/architecture.md`](docs/architecture.md) (full design rationale).
@@ -84,7 +85,7 @@ Each plugin is a directory under `plugins/`. Three component types, all auto-dis
 - **Skills** (`skills/<n>/SKILL.md`) — modular knowledge with progressive disclosure. Frontmatter: `name`, `description` (must include a recognized trigger phrase like "Use when …"). Supporting material in `references/`, templates in `assets/`.
 - **Commands** (`commands/<n>.md`) — slash commands. Frontmatter: `description`, `argument-hint`.
 
-Full conventions in [`docs/authoring.md`](docs/authoring.md). Authoring for portability across all five harnesses is the main concern; the adapter framework handles per-harness mechanics.
+Full conventions in [`docs/authoring.md`](docs/authoring.md). Authoring for portability across all six harnesses is the main concern; the adapter framework handles per-harness mechanics.
 
 ## Model tiers
 

CONTRIBUTING.md

@@ -1,8 +1,8 @@
 # Contributing to claude-agents
 
-Thanks for your interest in contributing. This marketplace ships to six agentic
-harnesses (Claude Code, OpenAI Codex CLI, Cursor, OpenCode, Gemini CLI, GitHub Copilot) from a single
-Markdown source.
+Thanks for your interest in contributing. This marketplace is authored for Claude Code
+and ships to six downstream harnesses (OpenAI Codex CLI, Cursor, OpenCode, Gemini CLI,
+GitHub Copilot, Antigravity CLI) from a single Markdown source.
 
 ## Start here
 
@@ -45,7 +45,7 @@ uv run ty check ../../tools/ src/plugin_eval/
 
 ## Cross-harness portability checklist
 
-Your content ships to five harnesses — some have stricter conventions than Claude Code:
+Your content ships to six harnesses — some have stricter conventions than Claude Code:
 
 - **Codex** hard-truncates skill bodies at 8 KB. Keep `SKILL.md` short; push detail
   into `references/details.md`.
@@ -54,6 +54,8 @@ Your content ships to five harnesses — some have stricter conventions than Cla
 - **Cursor** doesn't honor per-agent `tools:` allowlists — use it as a hint only.
 - **Copilot** maps Claude model aliases (`opus`/`sonnet`/`haiku`) to the GPT-5 family;
   agent `description` must be a plain string.
+- **Antigravity CLI** maps Claude model aliases to the Gemini 2.5 family; all tool names
+  are lowercase; agents use `agent.json` format with Gemini subagent spec.
 - All harnesses use ≤150-line context files. Don't bloat `AGENTS.md` / `CLAUDE.md`.
 
 `plugin-eval`'s `harness_portability` dimension catches most of these mechanically;

README.md

@@ -2,12 +2,12 @@
 
 > Production-ready agentic workflow building blocks: **83 plugins**, **191 agents**,
 > **155 skills**, **102 commands** — built for Claude Code and consumed natively by
-> OpenAI Codex CLI, Cursor, OpenCode, Gemini CLI, and GitHub Copilot from a single Markdown source.
+> OpenAI Codex CLI, Cursor, OpenCode, Gemini CLI, GitHub Copilot, and Antigravity CLI from a single Markdown source.
 
-[![Claude Code](https://img.shields.io/badge/Claude%20Code-native-blueviolet)](#claude-code) [![Codex CLI](https://img.shields.io/badge/Codex%20CLI-supported-black)](docs/harnesses.md) [![Cursor](https://img.shields.io/badge/Cursor-supported-purple)](docs/harnesses.md) [![OpenCode](https://img.shields.io/badge/OpenCode-supported-green)](docs/harnesses.md) [![Gemini CLI](https://img.shields.io/badge/Gemini%20CLI-supported-blue)](GEMINI.md) [![Copilot](https://img.shields.io/badge/Copilot-supported-lightgrey)](docs/harnesses.md)
+[![Claude Code](https://img.shields.io/badge/Claude%20Code-native-blueviolet)](#claude-code) [![Codex CLI](https://img.shields.io/badge/Codex%20CLI-supported-black)](docs/harnesses.md) [![Cursor](https://img.shields.io/badge/Cursor-supported-purple)](docs/harnesses.md) [![OpenCode](https://img.shields.io/badge/OpenCode-supported-green)](docs/harnesses.md) [![Gemini CLI](https://img.shields.io/badge/Gemini%20CLI-supported-blue)](GEMINI.md) [![Copilot](https://img.shields.io/badge/Copilot-supported-lightgrey)](docs/harnesses.md) [![Antigravity CLI](https://img.shields.io/badge/Antigravity%20CLI-supported-orange)](docs/harnesses.md)
 
 > [!NOTE]
-> One source-of-truth (`plugins/`), five harnesses. Each harness gets idiomatic,
+> One source-of-truth (`plugins/`), six harnesses. Each harness gets idiomatic,
 > harness-native artifacts — not lowest-common-denominator translations.
 > See [docs/harnesses.md](docs/harnesses.md) for the capability matrix.
 
@@ -24,12 +24,12 @@ Pick your harness:
 
 [→ Full Claude Code setup, troubleshooting, and plugin catalog](docs/usage.md)
 
-### Codex CLI · Cursor · OpenCode · Gemini CLI · Copilot
+### Codex CLI · Cursor · OpenCode · Gemini CLI · Copilot · Antigravity CLI
 
 ```bash
 gh repo clone wshobson/agents ~/agents
 cd ~/agents
-make generate HARNESS=<codex|cursor|opencode|gemini|copilot>
+make generate HARNESS=<codex|cursor|opencode|gemini|copilot|antigravity>
 ```
 
 Setup details and per-harness gotchas: [docs/harnesses.md](docs/harnesses.md). Gemini-specific setup: [GEMINI.md](GEMINI.md) (also auto-loaded by Gemini CLI).
@@ -73,7 +73,7 @@ Three-tier model strategy:
 
 ## Multi-harness support
 
-This marketplace ships to five agentic harnesses from one Markdown source. Each adapter
+This marketplace ships to six agentic harnesses from one Markdown source. Each adapter
 emits harness-native artifacts (not lowest-common-denominator translations):
 
 | Harness | Generates | Notes |
@@ -84,9 +84,10 @@ emits harness-native artifacts (not lowest-common-denominator translations):
 | **OpenCode** | `.opencode/agents/`, `.opencode/commands/`, `.opencode/skills/` | `permission:` block from `tools:` allowlist; OpenCode-safe skill names |
 | **Gemini CLI** | `skills/`, `agents/`, `commands/` (TOML) | Native skills + subagents (April 2026 spec) |
 | **Copilot** | `.copilot/agents/`, `.copilot/skills/`, `.copilot/commands/` | Markdown agent profiles + SKILL.md skills + commands-as-skills; model maps to GPT-5 family |
+| **Antigravity CLI** | `.antigravity/agents/`, `.antigravity/skills/`, `.antigravity/commands/` | Lowercase agent.json per agent + SKILL.md skills + commands-as-skills; model maps to Gemini 2.5 family |
 
 ```bash
-make generate-all                        # all five
+make generate-all                        # all six
 make validate                            # structural checks
 make garden                              # drift / dead-link / cap detection
 ```
@@ -141,6 +142,7 @@ integrations for this marketplace's other supported harnesses.
 | OpenCode | [integrations/opencode-plugin](https://github.com/major7apps/pensyve/tree/main/integrations/opencode-plugin) |
 | Gemini CLI | `gemini extensions install https://github.com/major7apps/pensyve` |
 | Copilot | `.copilot/` in repo root or `~/.copilot/` via `make install-copilot` |
+| Antigravity CLI | `.antigravity/` in repo root or `~/.antigravity/` via `make install-antigravity` |
 
 ## License
 

docs/authoring.md

@@ -1,6 +1,6 @@
 # Authoring portable plugin content
 
-Plugin content in this repo ships to **five** harnesses: OpenAI Codex CLI, Cursor, OpenCode, Gemini CLI, and GitHub Copilot. Claude Code is the source-of-truth. The adapter framework handles per-harness
+Plugin content in this repo ships to **six** harnesses: OpenAI Codex CLI, Cursor, OpenCode, Gemini CLI, GitHub Copilot, and Antigravity CLI. Claude Code is the source-of-truth. The adapter framework handles per-harness
 mechanics (frontmatter rewrites, format transforms, output paths) so you author one set of
 markdown files. But content choices still affect portability — this guide tells you what to
 do, and what to avoid, so the work you do for Claude Code translates cleanly everywhere.
@@ -105,12 +105,12 @@ clean naming — pick distinct names for skill/command pairs within a plugin.
 
 ### Model aliases
 
-| Source field | Codex | Cursor | OpenCode | Gemini | Copilot |
-|---|---|---|---|---|---|---|
-| `model: opus` | `gpt-5` | `inherit` | `anthropic/claude-opus-4-7` | `gemini-2.5-pro` | `gpt-5` |
-| `model: sonnet` | `gpt-5-mini` | `inherit` | `anthropic/claude-sonnet-4-6` | `gemini-2.5-pro` | `gpt-5-mini` |
-| `model: haiku` | `gpt-5-nano` | `inherit` | `anthropic/claude-haiku-4-5-20251001` | `gemini-2.5-flash` | `gpt-5-nano` |
-| `model: inherit` | `gpt-5` | `inherit` | `anthropic/claude-sonnet-4-6` | `gemini-2.5-pro` | `gpt-5` |
+| Source field | Codex | Cursor | OpenCode | Gemini | Copilot | Antigravity CLI |
+|---|---|---|---|---|---|---|---|---|
+| `model: opus` | `gpt-5` | `inherit` | `anthropic/claude-opus-4-7` | `gemini-2.5-pro` | `gpt-5` | `gemini-2.5-pro` |
+| `model: sonnet` | `gpt-5-mini` | `inherit` | `anthropic/claude-sonnet-4-6` | `gemini-2.5-pro` | `gpt-5-mini` | `gemini-2.5-pro` |
+| `model: haiku` | `gpt-5-nano` | `inherit` | `anthropic/claude-haiku-4-5-20251001` | `gemini-2.5-flash` | `gpt-5-nano` | `gemini-2.5-flash` |
+| `model: inherit` | `gpt-5` | `inherit` | `anthropic/claude-sonnet-4-6` | `gemini-2.5-pro` | `gpt-5` | `gemini-2.5-pro` |
 
 The adapter handles mapping. The `BARE_MODEL_ALIAS` lint is informational — it just notes
 that the mapping is implicit. If you want explicit, use `inherit`.
@@ -140,7 +140,7 @@ Things that work in Claude Code but degrade across harnesses:
 | Hooks (`hooks:` frontmatter) | Only Claude Code and OpenCode (via TS plugins). |
 | `color:` on agents | Cosmetic; dropped everywhere except Claude Code. |
 | Per-agent tool allowlist | Honored only on Claude Code/Gemini/OpenCode. Cursor and Codex have coarser models. |
-| Slash commands | Codex converts to skills. Gemini transpiles to TOML. Copilot emits `.copilot/commands/` prompt files. |
+| Slash commands | Codex converts to skills. Gemini transpiles to TOML. Copilot emits `.copilot/commands/` prompt files. Antigravity emits `.antigravity/commands/` SKILL.md files. |
 | Marketplace registry | Only Claude Code and Cursor have one. Gemini installs by URL; Codex/OpenCode have no marketplace. |
 
 When you must use a feature with no equivalent, the `harness_portability` lint won't fire

docs/round-trip-results.md

@@ -14,6 +14,7 @@ load the generated artifacts and report what it found.
 | **Codex CLI** | 0.133.0 | ✅ pass (structural) | All 191 agent TOMLs parse via Python `tomllib`; AGENTS.md within budget (43 lines / 500 tokens) | Codex doctor surfaces no errors; deeper "did the model actually load the skill" requires interactive verification. |
 | **Cursor** | (editor-only) | n/a | n/a | No CLI; manual verification recipe below. |
 | **Copilot** | (structural) | ✅ pass | 191 agent profiles, 155 skills, 25 commands all validated | No CLI round-trip tool yet; structural validation via `make validate` passes. |
+| **Antigravity CLI** | (structural) | ✅ pass | 191 agent.json profiles, 155 skills, 25 commands all validated | No CLI round-trip tool yet; structural validation via `make validate` passes. |
 
 ## Issues surfaced and fixed during round-trip
 
@@ -100,6 +101,27 @@ make generate HARNESS=cursor
 # 6. Skills under .claude/skills/ should auto-trigger from descriptions
 ```
 
+### Antigravity CLI (no CLI round-trip yet)
+
+```bash
+# Generate
+make generate HARNESS=antigravity
+
+# Structural validation (parses every generated artifact)
+make validate
+
+# Verify artifact tree
+ls .antigravity/agents/   # 191 agent.json profiles
+ls .antigravity/skills/   # 155 skill dirs (each with SKILL.md)
+ls .antigravity/commands/ # 25 command SKILL.md files
+
+# Global install (optional)
+make install-antigravity   # symlinks .antigravity/ -> ~/.antigravity/
+```
+
+Antigravity CLI currently lacks a CLI verification tool. Manual testing requires
+opening the Antigravity editor and verifying agents and skills load correctly.
+
 ### Copilot (no CLI round-trip yet)
 
 ```bash
@@ -128,7 +150,7 @@ The `tools/validate_generated.py` script approximates round-trip without install
 harnesses:
 
 ```bash
-make validate                 # all five harnesses
+make validate                 # all six harnesses
 make validate HARNESS=codex   # one only
 ```
 
@@ -158,6 +180,8 @@ the artifacts at runtime. Specifically untested by the automated suite:
   prompt.
 - Whether Copilot's agent profile and skill discovery actually loads our artifacts
   end-to-end (no CLI; requires VS Code editor).
+- Whether Antigravity CLI's agent.json and skill discovery actually loads our artifacts
+  end-to-end (no CLI; requires Antigravity editor).
 
 These require interactive use and API-token-burning runs. The recipes above show how
 to perform them manually.