garrytan · KiDDarn · Jul 12, 2026 · Jul 12, 2026 · Jul 12, 2026 · Jul 12, 2026
diff --git a/.gitignore b/.gitignore
@@ -23,6 +23,7 @@ bin/gstack-global-discover*
 .hermes/
 .gbrain/
 .gbrain-source
+.gemini/
 .context/
 extension/.auth.json
 # xterm assets are vendored from npm at build time; not source-of-truth.

diff --git a/AGENTS.md b/AGENTS.md
@@ -62,7 +62,7 @@ Invoke them by name (e.g., `/office-hours`).
 | `/retro` | Weekly retro with per-person breakdowns and shipping streaks. |
 | `/health` | Code quality dashboard (type checker, linter, tests, dead code). |
 | `/benchmark` | Performance regression detection (page load, Core Web Vitals). |
-| `/benchmark-models` | Cross-model benchmark for skills (Claude, GPT, Gemini side-by-side). |
+| `/benchmark-models` | Cross-model benchmark for skills (Claude, GPT, Gemini, and Agy side-by-side). |
 | `/cso` | OWASP Top 10 + STRIDE security audit. |
 | `/setup-gbrain` | Set up gbrain for cross-machine session memory sync. |
 | `/sync-gbrain` | Keep gbrain current with this repo's code; refresh agent search guidance in CLAUDE.md. |

diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,5 +1,27 @@
 # Changelog
 
+## [1.61.0.0] - 2026-07-12
+
+## **Permanent Antigravity (Agy) integration and benchmark support are live.**
+## **gstack now runs on Agy, registers/updates plugins automatically, and supports Agy model benchmarks.**
+
+Agy integration is now permanent and first-class. Running setup with `--host agy` generates and imports Agy-compatible skills, excludes incompatible skills, and registers the gstack plugin in Agy's plugin configuration. This release also fixes several test-suite and helper issues that were causing CI failures, and adds Agy model benchmarking.
+
+### Itemized changes
+
+#### Added
+
+- `hosts/agy.ts`: Declarative host configuration for Agy.
+- `test/helpers/providers/agy.ts`: Agy benchmark runner adapter that executes `agy --print` with safely skipped permissions.
+- `docs/AGY.md`: User documentation for Agy installation and integration.
+
+#### Fixed
+
+- `setup`: Support `--host agy` flag, build `plugin.json` and skill directories under `.gemini/config/plugins/gstack`, and trigger `agy plugin install`.
+- `bin/gstack-codex-session-import`: Added `-r` flag to `xargs` pipeline to prevent listing files in current directory when empty.
+- `test/gbrain-detect-install.test.ts`: Appended bun's executable directory to test environment `PATH` to resolve shebang resolution failures.
+- `test/gbrain-refresh-install-render.test.ts`: Resolved regex pattern mismatch in status check.
+
 ## [1.60.1.0] - 2026-07-09
 
 ## **The /autoplan dual-voice eval is back on the board, catching real regressions.**

diff --git a/CLAUDE.md b/CLAUDE.md
@@ -95,7 +95,7 @@ gstack/
 │   ├── claude.ts    # Primary host config
 │   ├── codex.ts, factory.ts, kiro.ts  # Existing hosts
 │   ├── opencode.ts, slate.ts, cursor.ts, openclaw.ts  # IDE hosts
-│   ├── hermes.ts, gbrain.ts  # Agent runtime hosts
+│   ├── hermes.ts, gbrain.ts, agy.ts  # Agent runtime hosts
 │   └── index.ts     # Registry: exports all, derives Host type
 ├── scripts/         # Build + DX tooling
 │   ├── gen-skill-docs.ts  # Template → SKILL.md generator (config-driven)

diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
@@ -308,11 +308,11 @@ worth the review overhead.
 
 ## Multi-host development
 
-gstack generates SKILL.md files for 8 hosts from one set of `.tmpl` templates.
+gstack generates SKILL.md files for 11 hosts from one set of `.tmpl` templates.
 Each host is a typed config in `hosts/*.ts`. The generator reads these configs
 to produce host-appropriate output (different frontmatter, paths, tool names).
 
-**Supported hosts:** Claude (primary), Codex, Factory, Kiro, OpenCode, Slate, Cursor, OpenClaw.
+**Supported hosts:** Claude (primary), Codex, Factory, Kiro, OpenCode, Slate, Cursor, OpenClaw, Hermes, GBrain, and Agy.
 
 ### Generating for all hosts
 
@@ -321,7 +321,8 @@ to produce host-appropriate output (different frontmatter, paths, tool names).
 bun run gen:skill-docs                    # Claude (default)
 bun run gen:skill-docs --host codex       # Codex
 bun run gen:skill-docs --host opencode    # OpenCode
-bun run gen:skill-docs --host all         # All 8 hosts
+bun run gen:skill-docs --host agy         # Google Antigravity
+bun run gen:skill-docs --host all         # All 11 hosts
 
 # Or use build, which does all hosts + compiles binaries
 bun run build

diff --git a/README.md b/README.md
@@ -101,7 +101,7 @@ These are conversational skills. Your OpenClaw agent runs them directly via chat
 
 ### Other AI Agents
 
-gstack works on 10 AI coding agents, not just Claude. Setup auto-detects which
+gstack works on 11 AI coding agents, not just Claude. Setup auto-detects which
 agents you have installed:
 
 ```bash
@@ -121,10 +121,13 @@ Or target a specific agent with `./setup --host <name>`:
 | Kiro | `--host kiro` | `~/.kiro/skills/gstack-*/` |
 | Hermes | `--host hermes` | `~/.hermes/skills/gstack-*/` |
 | GBrain (mod) | `--host gbrain` | `~/.gbrain/skills/gstack-*/` |
+| Google Antigravity (Agy) | `--host agy` | `~/.gemini/config/plugins/gstack/skills/` |
 
 **Want to add support for another agent?** See [docs/ADDING_A_HOST.md](docs/ADDING_A_HOST.md).
 It's one TypeScript config file, zero code changes.
 
+For Agy plugin registration and model benchmarking, see the [Agy integration guide](docs/AGY.md).
+
 ## See it work
 
 ```
@@ -241,7 +244,7 @@ Beyond the slash-command skills, gstack ships standalone CLIs for workflows that
 
 | Command | What it does |
 |---------|-------------|
-| `gstack-model-benchmark` | **Cross-model benchmark** — run the same prompt through Claude, GPT (via Codex CLI), and Gemini; compare latency, tokens, cost, and (optionally) LLM-judge quality score. Auth detected per provider, unavailable providers skip cleanly. Output as table, JSON, or markdown. `--dry-run` validates flags + auth without spending API calls. |
+| `gstack-model-benchmark` | **Cross-model benchmark** — run the same prompt through Claude, GPT (via Codex CLI), Gemini, and Agy; compare latency, tokens, cost, and (optionally) LLM-judge quality score. Auth detected per provider, unavailable providers skip cleanly. Output as table, JSON, or markdown. `--dry-run` validates flags + auth without spending API calls. See the [Agy integration guide](docs/AGY.md) for Agy setup. |
 | `gstack-taste-update` | **Design taste learning** — writes approvals and rejections from `/design-shotgun` into a persistent per-project taste profile. Decays 5%/week. Feeds back into future variant generation so the system learns what you actually pick. |
 | `gstack-ios-qa-daemon` | **iOS QA daemon** — Mac-side broker between an agent and a connected iPhone over USB CoreDevice. Loopback by default; `--tailnet` opens a Tailscale-facing listener with identity-gated capability tiers. Single-instance via flock on `~/.gstack/ios-qa-daemon.pid`. See [docs/howto-ios-testing-with-gstack.md](docs/howto-ios-testing-with-gstack.md). |
 | `gstack-ios-qa-mint` | **iOS allowlist manager** — owner-grant CLI for the tailnet allowlist. `grant`/`revoke`/`list` against `~/.gstack/ios-qa-allowlist.json` (mode 0600). Remote agents never auto-allowlist; this is the explicit-intent path. |

diff --git a/VERSION b/VERSION
@@ -1 +1 @@
-1.60.1.0
+1.61.0.0
diff --git a/benchmark-models/SKILL.md b/benchmark-models/SKILL.md
@@ -5,7 +5,7 @@ version: 1.0.0
 description: Cross-model benchmark for gstack skills. (gstack)
 triggers:
   - cross model benchmark
-  - compare claude gpt gemini
+  - compare claude gpt gemini agy
   - benchmark skill across models
   - which model should I use
 allowed-tools:
@@ -20,7 +20,7 @@ allowed-tools:
 ## When to invoke this skill
 
 Runs the same prompt through Claude,
-GPT (via Codex CLI), and Gemini side-by-side — compares latency, tokens, cost,
+GPT (via Codex CLI), Gemini, and Agy side-by-side — compares latency, tokens, cost,
 and optionally quality via LLM judge. Answers "which model is actually best
 for this skill?" with data instead of vibes. Separate from /benchmark, which
 measures web page performance. Use when: "benchmark models", "compare models",
@@ -578,12 +578,12 @@ If C: ask for the path. Verify it exists. Use as positional argument.
 ## Step 2: Choose providers
 
 ```bash
-"$BIN" --prompt "unused, dry-run" --models claude,gpt,gemini --dry-run
+"$BIN" --prompt "unused, dry-run" --models claude,gpt,gemini,agy --dry-run
 ```
 
 Show the dry-run output. The "Adapter availability" section tells the user which providers will actually run (OK) vs skip (NOT READY — remediation hint included).
 
-If ALL three show NOT READY: stop with a clear message — benchmark can't run without at least one authed provider. Suggest `claude login`, `codex login`, or `gemini login` / `export GOOGLE_API_KEY`.
+If ALL four show NOT READY: stop with a clear message — benchmark can't run without at least one authenticated provider. Suggest `claude login`, `codex login`, `gemini login` / `export GOOGLE_API_KEY`, or logging in through `agy`.
 
 If at least one is OK: AskUserQuestion:
 - **Simplify:** "Which models should we include? The dry-run above showed which are authed. Unauthed ones will be skipped cleanly — they won't abort the batch."

diff --git a/benchmark-models/SKILL.md.tmpl b/benchmark-models/SKILL.md.tmpl
@@ -4,7 +4,7 @@ preamble-tier: 1
 version: 1.0.0
 description: |
   Cross-model benchmark for gstack skills. Runs the same prompt through Claude,
-  GPT (via Codex CLI), and Gemini side-by-side — compares latency, tokens, cost,
+  GPT (via Codex CLI), Gemini, and Agy side-by-side — compares latency, tokens, cost,
   and optionally quality via LLM judge. Answers "which model is actually best
   for this skill?" with data instead of vibes. Separate from /benchmark, which
   measures web page performance. Use when: "benchmark models", "compare models",
@@ -15,7 +15,7 @@ voice-triggers:
   - "which model is best"
 triggers:
   - cross model benchmark
-  - compare claude gpt gemini
+  - compare claude gpt gemini agy
   - benchmark skill across models
   - which model should I use
 allowed-tools:
@@ -69,12 +69,12 @@ If C: ask for the path. Verify it exists. Use as positional argument.
 ## Step 2: Choose providers
 
 ```bash
-"$BIN" --prompt "unused, dry-run" --models claude,gpt,gemini --dry-run
+"$BIN" --prompt "unused, dry-run" --models claude,gpt,gemini,agy --dry-run
 ```
 
 Show the dry-run output. The "Adapter availability" section tells the user which providers will actually run (OK) vs skip (NOT READY — remediation hint included).
 
-If ALL three show NOT READY: stop with a clear message — benchmark can't run without at least one authed provider. Suggest `claude login`, `codex login`, or `gemini login` / `export GOOGLE_API_KEY`.
+If ALL four show NOT READY: stop with a clear message — benchmark can't run without at least one authenticated provider. Suggest `claude login`, `codex login`, `gemini login` / `export GOOGLE_API_KEY`, or logging in through `agy`.
 
 If at least one is OK: AskUserQuestion:
 - **Simplify:** "Which models should we include? The dry-run above showed which are authed. Unauthed ones will be skipped cleanly — they won't abort the batch."

diff --git a/bin/gstack-codex-session-import b/bin/gstack-codex-session-import
@@ -65,7 +65,7 @@ case "$MODE" in
       exit 0
     fi
     LATEST=$(find "$CODEX_SESSIONS_ROOT" -type f -name "rollout-*.jsonl" -print 2>/dev/null \
-      | xargs ls -t 2>/dev/null | head -1 || true)
+      | xargs -r ls -t 2>/dev/null | head -1 || true)
     if [ -z "$LATEST" ]; then
       echo "NO_SESSIONS: no rollout-*.jsonl files under $CODEX_SESSIONS_ROOT"
       exit 0

diff --git a/bin/gstack-model-benchmark b/bin/gstack-model-benchmark
@@ -31,11 +31,13 @@ import { runBenchmark, formatTable, formatJson, formatMarkdown, type BenchmarkIn
 import { ClaudeAdapter } from '../test/helpers/providers/claude';
 import { GptAdapter } from '../test/helpers/providers/gpt';
 import { GeminiAdapter } from '../test/helpers/providers/gemini';
+import { AgyAdapter } from '../test/helpers/providers/agy';
 
 const ADAPTER_FACTORIES = {
   claude: () => new ClaudeAdapter(),
   gpt: () => new GptAdapter(),
   gemini: () => new GeminiAdapter(),
+  agy: () => new AgyAdapter(),
 };
 
 type OutputFormat = 'table' | 'json' | 'markdown';
@@ -76,13 +78,13 @@ function positionalArgs(args: string[]): string[] {
   return positional;
 }
 
-function parseProviders(s: string | undefined): Array<'claude' | 'gpt' | 'gemini'> {
+function parseProviders(s: string | undefined): Array<'claude' | 'gpt' | 'gemini' | 'agy'> {
   if (!s) return ['claude'];
-  const seen = new Set<'claude' | 'gpt' | 'gemini'>();
+  const seen = new Set<'claude' | 'gpt' | 'gemini' | 'agy'>();
   for (const p of s.split(',').map(x => x.trim()).filter(Boolean)) {
-    if (p === 'claude' || p === 'gpt' || p === 'gemini') seen.add(p);
+    if (p === 'claude' || p === 'gpt' || p === 'gemini' || p === 'agy') seen.add(p);
     else {
-      console.error(`WARN: unknown provider '${p}' — skipping. Valid: claude, gpt, gemini.`);
+      console.error(`WARN: unknown provider '${p}' — skipping. Valid: claude, gpt, gemini, agy.`);
     }
   }
   return seen.size ? Array.from(seen) : ['claude'];
@@ -149,7 +151,7 @@ async function main(): Promise<void> {
 
 async function dryRunReport(opts: {
   prompt: string;
-  providers: Array<'claude' | 'gpt' | 'gemini'>;
+  providers: Array<'claude' | 'gpt' | 'gemini' | 'agy'>;
   workdir: string;
   timeoutMs: number;
   output: OutputFormat;

diff --git a/docs/ADDING_A_HOST.md b/docs/ADDING_A_HOST.md
@@ -1,7 +1,7 @@
 # Adding a New Host to gstack
 
 gstack uses a declarative host config system. Each supported AI coding agent
-(Claude, Codex, Factory, Kiro, OpenCode, Slate, Cursor, OpenClaw) is defined
+(Claude, Codex, Factory, Kiro, OpenCode, Slate, Cursor, OpenClaw, Hermes, GBrain, Agy) is defined
 as a typed TypeScript config object. Adding a new host means creating one file
 and re-exporting it. Zero code changes to the generator, setup, or tooling.
 
@@ -17,6 +17,9 @@ hosts/
 ├── slate.ts         # Slate (Random Labs)
 ├── cursor.ts        # Cursor
 ├── openclaw.ts      # OpenClaw (hybrid: config + adapter)
+├── hermes.ts        # Hermes
+├── gbrain.ts        # GBrain
+├── agy.ts           # Google Antigravity
 └── index.ts         # Registry: imports all, derives Host type
 ```
 

diff --git a/docs/AGY.md b/docs/AGY.md
@@ -0,0 +1,71 @@
+# Google Antigravity (Agy) Integration
+
+gstack supports Google Antigravity (Agy) as a first-class agent host. This integration allows you to generate and run Agy-compatible skills, load them dynamically via the `agy` plugin system, and benchmark agy models alongside Claude, OpenAI, and Gemini.
+
+## Features
+
+- **Automatic Setup**: Installs gstack skills directly into the Antigravity config directory via `./setup --host agy`.
+- **Agy Plugin Compatibility**: Packages skills as a standard Agy plugin with a `plugin.json` manifest, automatically registered with `agy plugin install`.
+- **Exclusion of Incompatible Skills**: Automatically filters out Claude-specific wrappers (like `gstack-codex`) from Agy generation.
+- **Benchmark Integration**: Register Agy as a benchmark provider using `agy --print` to trace latency, quality, and tokens.
+
+---
+
+## Setup & Installation
+
+To generate and install gstack skills into your local Antigravity environment:
+
+```bash
+./setup --host agy
+```
+
+### What Setup Does
+
+1. **Precompile & Prepare**: Ensures the browse tool and other runtime dependencies are compiled.
+2. **Generate Skills**: Runs `bun run gen:skill-docs --host agy` to generate Agy-compatible frontmatter.
+3. **Initialize Plugin**: Creates the plugin directory structure under `.gemini/config/plugins/gstack`.
+4. **Symlink/Copy Assets**: Places necessary executables (`bin`, `browse`, `review`, `qa`) into the plugin root.
+5. **Register Plugin**: Invokes `agy plugin install` to import the gstack plugin into `~/.gemini/config/import_manifest.json`.
+
+---
+
+## Plugin Structure
+
+When installed, the gstack plugin resides at:
+`~/.gemini/config/plugins/gstack/`
+
+The directory layout matches the standard Agy plugin structure:
+
+```
+~/.gemini/config/plugins/gstack/
+├── plugin.json         # Manifest declaring name, version, and license
+├── skills/             # Generated Agy-compatible skills
+│   ├── gstack/         # Root gstack skill (/gstack)
+│   ├── gstack-qa/      # QA testing skill (/qa)
+│   └── ...
+├── bin/                # Shared executables and CLI tools
+└── browse/             # Browse tool distribution files
+```
+
+---
+
+## Benchmarking with Agy
+
+Agy is registered as a model benchmark provider. You can execute prompts against Agy using the gstack benchmark runner.
+
+### Pre-requisites
+- The `agy` CLI must be installed and available in your `PATH`.
+- Credentials must be configured in `~/.gemini/oauth_creds.json`.
+
+### Run Benchmark
+To run a dry run and check availability:
+```bash
+gstack-model-benchmark --models agy --dry-run
+```
+
+To execute a benchmark run:
+```bash
+gstack-model-benchmark --models agy --prompt "Explain quantum computing in one sentence."
+```
+
+The runner automatically passes `--dangerously-skip-permissions` to the CLI to run headless without hanging on interactive permission prompts.
diff --git a/hosts/agy.ts b/hosts/agy.ts
@@ -0,0 +1,59 @@
+import type { HostConfig } from '../scripts/host-config';
+
+const agy: HostConfig = {
+  name: 'agy',
+  displayName: 'Antigravity',
+  cliCommand: 'agy',
+  cliAliases: [],
+
+  globalRoot: '.gemini/config/plugins/gstack/skills',
+  localSkillRoot: '.gemini/config/plugins/gstack/skills',
+  hostSubdir: '.gemini/config/plugins/gstack',
+  usesEnvVars: true,
+
+  frontmatter: {
+    mode: 'allowlist',
+    keepFields: [
+      'name',
+      'preamble-tier',
+      'version',
+      'description',
+      'allowed-tools',
+      'triggers',
+      'hooks',
+      'gbrain',
+    ],
+    descriptionLimit: null,
+  },
+
+  generation: {
+    generateMetadata: false,
+    skipSkills: ['codex'],
+  },
+
+  pathRewrites: [
+    { from: '~/.claude/skills/gstack', to: '$GSTACK_ROOT' },
+    { from: '.claude/skills/gstack', to: '.gemini/config/plugins/gstack/skills' },
+    { from: '.claude/skills', to: '.gemini/config/plugins/gstack' },
+  ],
+
+  toolRewrites: {},
+
+  suppressedResolvers: ['GBRAIN_CONTEXT_LOAD', 'GBRAIN_SAVE_RESULTS'],
+
+  runtimeRoot: {
+    globalSymlinks: ['bin', 'browse/dist', 'browse/bin', 'gstack-upgrade', 'ETHOS.md'],
+    globalFiles: {
+      'review': ['checklist.md', 'TODOS-format.md'],
+    },
+  },
+
+  install: {
+    prefixable: false,
+    linkingStrategy: 'symlink-generated',
+  },
+
+  learningsMode: 'basic',
+};
+
+export default agy;
diff --git a/hosts/index.ts b/hosts/index.ts
@@ -16,9 +16,10 @@ import cursor from './cursor';
 import openclaw from './openclaw';
 import hermes from './hermes';
 import gbrain from './gbrain';
+import agy from './agy';
 
 /** All registered host configs. Add new hosts here. */
-export const ALL_HOST_CONFIGS: HostConfig[] = [claude, codex, factory, kiro, opencode, slate, cursor, openclaw, hermes, gbrain];
+export const ALL_HOST_CONFIGS: HostConfig[] = [claude, codex, factory, kiro, opencode, slate, cursor, openclaw, hermes, gbrain, agy];
 
 /** Map from host name to config. */
 export const HOST_CONFIG_MAP: Record<string, HostConfig> = Object.fromEntries(
@@ -65,4 +66,4 @@ export function getExternalHosts(): HostConfig[] {
 }
 
 // Re-export individual configs for direct import
-export { claude, codex, factory, kiro, opencode, slate, cursor, openclaw, hermes, gbrain };
+export { claude, codex, factory, kiro, opencode, slate, cursor, openclaw, hermes, gbrain, agy };