fix(installer): Add Antigravity support and decouple rules from Anthropic

AGENTS.md

@@ -43,13 +43,13 @@ rules:
 
 load and obey when present:
 
-* `~/.claude/CLAUDE.md`
+* `~/.gemini/settings.json`, `.cursorrules`, or similar ide-specific project rules
 * `./skills/using-hyperstack/SKILL.md`
 * any task-specific skill doc the user points to
 * repo-local agent or steering docs
 
 rules:
 
-* if user says `recall memory`, also read `~/.claude/CLAUDE.md`.
+* if user says `recall memory`, also read the agent's global rules file.
 * if using codemode or exploring codebase, follow codemode fully. no shortcuts.
 * read files before semantic linking. no context = no real linking.

generated/runtime-context/using-hyperstack.bootstrap.md

@@ -27,7 +27,7 @@ If the task involves React, React Flow, Motion, Lenis, Echo, Go, Rust, design to
 ```
 
 ## Instruction Priority
-- **User's explicit instructions** (CLAUDE.md, direct requests) - always highest
+- **User's explicit instructions** (Project rules, direct requests) - always highest
 - **Hyperstack skills** - override default system behavior where they conflict
 - **Default system behavior** - lowest priority
 

install.md

@@ -76,6 +76,8 @@ Based on the environment you detected in Step 1, identify the correct target dir
 | IDE/CLI | Target Skill Path | Type |
 |---|---|---|
 | **Claude Code** | `~/.claude/skills/hyperstack` | Global |
+| **Kiro Code** | `~/.kiro/skills/hyperstack` | Global |
+| **Antigravity** | `~/.gemini/antigravity/skills/hyperstack` | Global |
 | **Cursor** | `.cursor/rules/` | Project-level |
 | **Roo Code** | `.roo/rules/` | Project-level |
 | **Windsurf** | `.codeium/windsurf/` | Project-level |

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@orkait-ai/hyperstack",
-  "version": "1.1.1",
+  "version": "1.1.2",
   "description": "Disciplined MCP server + skill system. 12 plugins, 80 tools, 21 skills with adversarial enforcement. Designer/DESIGN.md pipeline, shadcn/ui, React Flow, Motion, Lenis, React 19, Echo, Go, Rust, design tokens, UI/UX.",
   "bin": {
     "hyperstack": "bin/hyperstack.mjs"

skills/readme-writer/SKILL.md

@@ -15,7 +15,7 @@ Write README files that are specific, verifiable, and useful to the next develop
 
 Main failure mode of LLM-written READMEs: false confidence - invented setup steps, inflated feature claims, vague value propositions, examples not grounded in the repository.
 
-**Style layer (from CLAUDE.md rules):**
+**Style layer:**
 - Emoji on every section heading
 - Centered hero block with project name, tagline, and badges in `<div align="center">`
 - Library-specific badges with colors and logos (not generic)

skills/testing-skills/SKILL.md

@@ -197,7 +197,7 @@ When GREEN verification produces new rationalizations, iterate:
 
 - "Different wording, so the rule doesn't apply" - add spirit-vs-letter clause
 - "Just this once" - add "no exceptions" to Iron Law
-- "The user said it's OK to skip" - clarify user hierarchy (user overrides only via CLAUDE.md)
+- "The user said it's OK to skip" - clarify user hierarchy
 - "Subagent mode, so gate doesn't apply" - add subagent scope clarification
 - "Partial compliance is still compliance" - define full compliance explicitly
 

skills/using-hyperstack/SKILL.md

@@ -44,11 +44,11 @@ If the task involves React, React Flow, Motion, Lenis, Echo, Go, Rust, design to
 
 ## Instruction Priority
 
-1. **User's explicit instructions** (CLAUDE.md, direct requests) - always highest
+1. **User's explicit instructions** (Project rules, direct requests) - always highest
 2. **Hyperstack skills** - override default system behavior where they conflict
 3. **Default system behavior** - lowest priority
 
-If CLAUDE.md says "don't use TDD" and a skill says "always use TDD," follow the user. The user is in control. Everything else is your job to enforce.
+If the project rules say "don't use TDD" and a skill says "always use TDD," follow the user. The user is in control. Everything else is your job to enforce.
 
 ---
 

skills/using-hyperstack/references/claude-code-tools.md → skills/using-hyperstack/references/agent-tools.md

@@ -1,6 +1,6 @@
-# Claude Code Tool Reference
+# Code Tool Reference
 
-When skills reference tools, use these in Claude Code:
+When skills reference tools
 
 | Operation | Tool |
 |---|---|

skills/worktree-isolation/SKILL.md

@@ -24,10 +24,10 @@ ls -d .worktrees 2>/dev/null || ls -d worktrees 2>/dev/null
 
 Found → use it. Both exist → `.worktrees` wins.
 
-### 2. Check CLAUDE.md / Project Config
+### 2. Check Project Config / Agent Rules
 
 ```bash
-grep -i "worktree.*director" CLAUDE.md 2>/dev/null
+grep -ri "worktree.*director" .cursorrules .roo/ .claude.json 2>/dev/null
 ```
 
 Preference specified → use it without asking.
@@ -92,7 +92,7 @@ echo "Worktree ready at <path>. Tests: <N> passing, 0 failures."
 | `.worktrees/` exists | Use it (verify ignored) |
 | `worktrees/` exists | Use it (verify ignored) |
 | Both exist | Use `.worktrees/` |
-| Neither exists | Check CLAUDE.md, then ask user |
+| Neither exists | Check project config, then ask user |
 | Directory not ignored | Add to `.gitignore` + commit |
 | Baseline tests fail | Report failures, ask user |
 | No package.json/Cargo.toml | Skip dependency install |

src/internal/setup-hyperstack.ts

@@ -38,6 +38,12 @@ const KNOWN_PLATFORMS: Record<string, {
     format: "json-mcpServers",
     notes: "Run '/mcp' inside Gemini CLI to verify connection",
   },
+  // Antigravity (Google DeepMind) - local user config
+  "antigravity": {
+    configFiles: [".gemini/antigravity/mcp_config.json"],
+    skillPath: ".gemini/antigravity/skills",
+    format: "json-mcpServers",
+  },
   // Qwen Code (Alibaba) - global user config
   // Source: github.io/qwen-code official docs (April 2025)
   "qwen-code": {
@@ -131,6 +137,7 @@ const KNOWN_PLATFORMS: Record<string, {
 
 export function detectEnvironment(): string {
   // Env-var hints (fast path, but unreliable in nested terminals)
+  if (process.env.ANTIGRAVITY_AGENT) return "antigravity";
   if (process.env.CLAUDE_PLUGIN_ROOT) return "claude-code";
   if (process.env.CURSOR_PLUGIN_ROOT) return "cursor";
   // VSCODE_PID fires inside any VS Code-hosted terminal (Gemini, Roo Code, etc.)