chore: expose agent rules and skills to agy via AGENTS.md

.agents/skills

@@ -0,0 +1 @@
+../.claude/skills

.github/copilot-instructions.md

@@ -322,7 +322,7 @@ uv run alembic upgrade head
 
 - **Documentation**: Check `docs/` directory for architecture, workflow, and development guides
 - **Examples**: Look at existing plot implementations in `plots/` for patterns
-- **CLAUDE.md**: Additional guidance specific to Claude AI agent workflows
+- **CLAUDE.md**: Additional guidance specific to Claude AI agent workflows; `AGENTS.md` is a symlink to it, so agents that read `AGENTS.md` get the same rules
 - **Issue Templates**: Use GitHub issue templates for consistent problem reporting
 
 ### Common Issues

AGENTS.md

@@ -0,0 +1 @@
+CLAUDE.md

CLAUDE.md

@@ -93,6 +93,7 @@ Known gaps with NO verification loop yet (reason through carefully and say so in
 ## Claude Code Configuration
 
 - **Commands directory**: Commands live in `agentic/commands/` (agent-agnostic). A symlink `.claude/commands/ → ../agentic/commands/` ensures Claude Code slash-command resolution works. Do not create commands directly in `.claude/commands/`.
+- **Other agents read the same files**: `AGENTS.md` is a symlink to this file and `.agents/skills/ → ../.claude/skills/` exposes the project skills, so the Antigravity CLI (`agy`), Codex, and other tools that read `AGENTS.md` or `.agents/skills/` load the same rules and skills. Edit `CLAUDE.md` and `.claude/skills/`, never the link paths, and never replace either link with a real file or directory. Rules worded for Claude Code tools (Edit/Write, `model: opus`, subagents, MCP servers) apply to other agents by analogy: their file-editing tool, their strongest model tier, their delegation mechanism.
 
 ## CRITICAL: Mandatory Workflow for New Specs and Implementations
 

agentic/docs/project-guide.md

@@ -206,6 +206,7 @@ Example: `plots/scatter-basic/` contains everything for the basic scatter plot.
 - **`agentic/`**: AI workflow layer (composable phases, prompt templates, runtime state)
   - **`agentic/workflows/`**: Click CLI scripts (plan, build, test, review + orchestrators)
   - **`agentic/commands/`**: Markdown prompt templates
+- **Agent instruction files**: `CLAUDE.md` holds the rules and `.claude/skills/` the skills. Three symlinks make them visible to other tools without a copy to keep in sync: `.claude/commands/` points to `agentic/commands/`, `AGENTS.md` points to `CLAUDE.md` (read by the Antigravity CLI `agy`, Codex, and the GitHub Copilot coding agent), and `.agents/skills/` points to `.claude/skills/` (the skills directory `agy` reads). `.github/copilot-instructions.md` is the separate, hand-synced guide for Copilot. `tests/unit/test_agent_instructions.py` fails if a link is replaced by a real file or points elsewhere.
 - **`automation/`**: CI/CD helper scripts (workflow_cli, label_manager, sync_to_postgres)
 - **`infra/`**: Infrastructure that would otherwise live only in a dashboard
   - **`infra/cloudflare/`**: Source of the apex `anyplot.ai/api/*` Worker, plus the origin gate's rollout and measuring procedure

changelog.d/agents-md-bridge.md

@@ -0,0 +1,8 @@
+### Added
+
+- **Agent rules and skills for non-Claude agents.** `AGENTS.md` is a symlink to
+  `CLAUDE.md` and `.agents/skills/` a symlink to `.claude/skills/`, so the
+  Antigravity CLI (`agy`), Codex, and other tools that read those paths load the
+  same rules and skills from one source instead of a third file to keep in sync.
+  A unit test fails if either link is replaced by a real file or points
+  elsewhere. (#12187)

tests/unit/test_agent_instructions.py

@@ -235,6 +235,29 @@ def test_the_commands_symlink_points_where_it_says() -> None:
     assert link.resolve() == (REPO_ROOT / "agentic" / "commands").resolve()
 
 
+@pytest.mark.parametrize(
+    ("link", "target", "raw"),
+    [
+        (REPO_ROOT / "AGENTS.md", CLAUDE_MD, "CLAUDE.md"),
+        (REPO_ROOT / ".agents" / "skills", REPO_ROOT / ".claude" / "skills", "../.claude/skills"),
+    ],
+    ids=["AGENTS.md", ".agents-skills"],
+)
+def test_the_agents_bridge_links_point_at_the_claude_sources(link: Path, target: Path, raw: str) -> None:
+    """`AGENTS.md` and `.agents/skills/` are how agy, Codex and other non-Claude
+    agents find the rules and skills (CLAUDE.md, "Claude Code Configuration").
+
+    Same failure as the commands link: a real file or directory in their place
+    still exists and still reads plausibly, but it is a third copy nobody keeps
+    in sync. The raw link text is pinned as well, because a relative link is
+    what keeps it valid in every checkout and worktree; an absolute one resolves
+    on the machine that made it and nowhere else.
+    """
+    assert link.is_symlink(), f"{link} is documented as a symlink"
+    assert link.readlink() == Path(raw), f"{link} should be the relative link {raw}"
+    assert link.resolve() == target.resolve()
+
+
 def test_every_named_skill_exists() -> None:
     """The routing table in CLAUDE.md sends work to `.claude/skills/<name>/`.