wafer-cli 0.2.24__tar.gz → 0.2.26__tar.gz

wafer_cli-0.2.26/PKG-INFO

@@ -0,0 +1,107 @@
+Metadata-Version: 2.4
+Name: wafer-cli
+Version: 0.2.26
+Summary: CLI for running GPU workloads, managing remote workspaces, and evaluating/optimizing kernels
+Requires-Python: >=3.11
+Description-Content-Type: text/markdown
+Requires-Dist: typer>=0.12.0
+Requires-Dist: trio>=0.24.0
+Requires-Dist: trio-asyncio>=0.15.0
+Requires-Dist: wafer-core>=0.1.0
+Requires-Dist: perfetto>=0.16.0
+Requires-Dist: posthog>=3.0.0
+Provides-Extra: dev
+Requires-Dist: pytest>=8.0.0; extra == "dev"
+Requires-Dist: pytest-cov>=4.1.0; extra == "dev"
+Requires-Dist: diff-cover>=8.0.0; extra == "dev"
+Requires-Dist: ruff>=0.4.0; extra == "dev"
+
+# Wafer CLI
+
+Wafer CLI gives coding agents direct access to GPU docs, trace analysis, and remote kernel evaluation.
+It helps you develop and optimize GPU kernels even when you are not working on a machine with a GPU.
+
+## Key features
+
+- Query GPU documentation with citations
+- Analyze GPU traces and profiles
+- Evaluate kernels on remote GPUs for correctness and performance
+- Run commands on GPU targets (remote or local)
+- Manage persistent workspaces
+
+## Quick start
+
+```bash
+uv tool install wafer-cli
+wafer login
+wafer remote-run -- nvidia-smi
+```
+
+## Common commands
+
+```bash
+wafer workspaces list
+wafer workspaces create my-workspace --wait
+wafer agent -t ask-docs --corpus cuda "What causes shared memory bank conflicts?"
+wafer agent -t trace-analyze --args trace=./profile.ncu-rep "Why is this kernel slow?"
+wafer evaluate --impl kernel.py --reference ref.py --test-cases tests.json --benchmark
+wafer nvidia ncu analyze profile.ncu-rep
+wafer corpus list
+```
+
+## Typical workflows
+
+### Query GPU documentation
+
+Download a documentation corpus and ask questions with citations.
+
+```bash
+wafer corpus download cuda
+wafer agent -t ask-docs --corpus cuda "What causes shared memory bank conflicts?"
+```
+
+### Analyze performance traces
+
+Use the trace analysis template or query trace data directly.
+
+```bash
+wafer agent -t trace-analyze --args trace=./profile.ncu-rep "Why is this kernel slow?"
+wafer nvidia perfetto query trace.json \
+  "SELECT name, dur/1e6 as ms FROM slice WHERE cat='kernel' ORDER BY dur DESC LIMIT 10"
+```
+
+### Evaluate kernels on remote GPUs
+
+Run correctness and performance checks on a remote target.
+
+```bash
+wafer evaluate \
+  --impl ./kernel.py \
+  --reference ./reference.py \
+  --test-cases ./tests.json \
+  --benchmark
+```
+
+### Run commands on a remote GPU
+
+```bash
+wafer remote-run -- nvidia-smi
+wafer remote-run --upload-dir ./my_code -- python3 train.py
+```
+
+### Manage workspaces
+
+```bash
+wafer workspaces list
+wafer workspaces create my-workspace --wait
+wafer workspaces ssh <workspace-id>
+wafer workspaces delete <workspace-id>
+```
+
+## Install the CLI skill (optional)
+
+```bash
+wafer skill install
+# or
+wafer skill install -t <claude/codex>
+```

wafer_cli-0.2.26/README.md

@@ -0,0 +1,89 @@
+# Wafer CLI
+
+Wafer CLI gives coding agents direct access to GPU docs, trace analysis, and remote kernel evaluation.
+It helps you develop and optimize GPU kernels even when you are not working on a machine with a GPU.
+
+## Key features
+
+- Query GPU documentation with citations
+- Analyze GPU traces and profiles
+- Evaluate kernels on remote GPUs for correctness and performance
+- Run commands on GPU targets (remote or local)
+- Manage persistent workspaces
+
+## Quick start
+
+```bash
+uv tool install wafer-cli
+wafer login
+wafer remote-run -- nvidia-smi
+```
+
+## Common commands
+
+```bash
+wafer workspaces list
+wafer workspaces create my-workspace --wait
+wafer agent -t ask-docs --corpus cuda "What causes shared memory bank conflicts?"
+wafer agent -t trace-analyze --args trace=./profile.ncu-rep "Why is this kernel slow?"
+wafer evaluate --impl kernel.py --reference ref.py --test-cases tests.json --benchmark
+wafer nvidia ncu analyze profile.ncu-rep
+wafer corpus list
+```
+
+## Typical workflows
+
+### Query GPU documentation
+
+Download a documentation corpus and ask questions with citations.
+
+```bash
+wafer corpus download cuda
+wafer agent -t ask-docs --corpus cuda "What causes shared memory bank conflicts?"
+```
+
+### Analyze performance traces
+
+Use the trace analysis template or query trace data directly.
+
+```bash
+wafer agent -t trace-analyze --args trace=./profile.ncu-rep "Why is this kernel slow?"
+wafer nvidia perfetto query trace.json \
+  "SELECT name, dur/1e6 as ms FROM slice WHERE cat='kernel' ORDER BY dur DESC LIMIT 10"
+```
+
+### Evaluate kernels on remote GPUs
+
+Run correctness and performance checks on a remote target.
+
+```bash
+wafer evaluate \
+  --impl ./kernel.py \
+  --reference ./reference.py \
+  --test-cases ./tests.json \
+  --benchmark
+```
+
+### Run commands on a remote GPU
+
+```bash
+wafer remote-run -- nvidia-smi
+wafer remote-run --upload-dir ./my_code -- python3 train.py
+```
+
+### Manage workspaces
+
+```bash
+wafer workspaces list
+wafer workspaces create my-workspace --wait
+wafer workspaces ssh <workspace-id>
+wafer workspaces delete <workspace-id>
+```
+
+## Install the CLI skill (optional)
+
+```bash
+wafer skill install
+# or
+wafer skill install -t <claude/codex>
+```

{wafer_cli-0.2.24 → wafer_cli-0.2.26}/pyproject.toml

@@ -1,7 +1,8 @@
 [project]
 name = "wafer-cli"
-version = "0.2.24"
-description = "CLI tool for running commands on remote GPUs and GPU kernel optimization agent"
+version = "0.2.26"
+description = "CLI for running GPU workloads, managing remote workspaces, and evaluating/optimizing kernels"
+readme = "README.md"
 requires-python = ">=3.11"
 dependencies = [
     "typer>=0.12.0",

{wafer_cli-0.2.24 → wafer_cli-0.2.26}/tests/test_analytics.py

@@ -467,7 +467,7 @@ class TestLoginLogoutAnalytics:
              patch("wafer.analytics.track_login") as mock_track_login, \
              patch("wafer.analytics.init_analytics", return_value=True):
 
-            runner.invoke(app, ["login", "--token", "test-token"])
+            runner.invoke(app, ["auth", "login", "--token", "test-token"])
 
             # track_login should be called
             mock_track_login.assert_called_once_with("test-user-id", "test@example.com")
@@ -484,7 +484,7 @@ class TestLoginLogoutAnalytics:
              patch("wafer.analytics.track_logout") as mock_track_logout, \
              patch("wafer.analytics.init_analytics", return_value=True):
 
-            result = runner.invoke(app, ["logout"])
+            result = runner.invoke(app, ["auth", "logout"])
 
             assert result.exit_code == 0
             mock_track_logout.assert_called_once()

{wafer_cli-0.2.24 → wafer_cli-0.2.26}/tests/test_billing.py

@@ -210,7 +210,7 @@ class TestBillingUsageCommand:
                 )
                 mock_client.return_value.__enter__.return_value.get.return_value = mock_response
 
-                result = runner.invoke(app, ["billing"])
+                result = runner.invoke(app, ["config", "billing"])
 
                 assert result.exit_code != 0
                 assert "login" in result.output.lower()
@@ -242,7 +242,7 @@ class TestBillingUsageCommand:
                     mock_response.raise_for_status.return_value = None
                     mock_client.return_value.__enter__.return_value.get.return_value = mock_response
 
-                    result = runner.invoke(app, ["billing", "--json"])
+                    result = runner.invoke(app, ["config", "billing", "--json"])
 
                     assert result.exit_code == 0
                     data = json.loads(result.stdout)
@@ -275,7 +275,7 @@ class TestBillingUsageCommand:
                     mock_response.raise_for_status.return_value = None
                     mock_client.return_value.__enter__.return_value.get.return_value = mock_response
 
-                    result = runner.invoke(app, ["billing"])
+                    result = runner.invoke(app, ["config", "billing"])
 
                     assert result.exit_code == 0
                     assert "Pro" in result.output
@@ -294,7 +294,7 @@ class TestBillingUsageCommand:
                         httpx.RequestError("Connection failed")
                     )
 
-                    result = runner.invoke(app, ["billing"])
+                    result = runner.invoke(app, ["config", "billing"])
 
                     assert result.exit_code != 0
                     assert "error" in result.output.lower() or "reach" in result.output.lower()
@@ -317,7 +317,7 @@ class TestBillingTopupCommand:
                 )
                 mock_client.return_value.__enter__.return_value.post.return_value = mock_response
 
-                result = runner.invoke(app, ["billing", "topup"])
+                result = runner.invoke(app, ["config", "billing", "topup"])
 
                 assert result.exit_code != 0
                 assert "login" in result.output.lower()
@@ -343,7 +343,7 @@ class TestBillingTopupCommand:
                     mock_client.return_value.__enter__.return_value.post.return_value = mock_response
 
                     with patch("webbrowser.open") as mock_browser:
-                        result = runner.invoke(app, ["billing", "topup"])
+                        result = runner.invoke(app, ["config", "billing", "topup"])
 
                         assert result.exit_code == 0
                         # Verify $25 = 2500 cents was sent
@@ -372,7 +372,7 @@ class TestBillingTopupCommand:
                     mock_client.return_value.__enter__.return_value.post.return_value = mock_response
 
                     with patch("webbrowser.open") as mock_browser:
-                        result = runner.invoke(app, ["billing", "topup", "100"])
+                        result = runner.invoke(app, ["config", "billing", "topup", "100"])
 
                         assert result.exit_code == 0
                         call_args = mock_client.return_value.__enter__.return_value.post.call_args
@@ -381,14 +381,14 @@ class TestBillingTopupCommand:
 
     def test_amount_below_minimum(self) -> None:
         """Amount below $10 should error."""
-        result = runner.invoke(app, ["billing", "topup", "5"])
+        result = runner.invoke(app, ["config", "billing", "topup", "5"])
 
         assert result.exit_code != 0
         assert "10" in result.output  # Should mention minimum
 
     def test_amount_above_maximum(self) -> None:
         """Amount above $500 should error."""
-        result = runner.invoke(app, ["billing", "topup", "600"])
+        result = runner.invoke(app, ["config", "billing", "topup", "600"])
 
         assert result.exit_code != 0
         assert "500" in result.output  # Should mention maximum
@@ -410,7 +410,7 @@ class TestBillingTopupCommand:
                     )
                     mock_client.return_value.__enter__.return_value.post.return_value = mock_response
 
-                    result = runner.invoke(app, ["billing", "topup"])
+                    result = runner.invoke(app, ["config", "billing", "topup"])
 
                     assert result.exit_code != 0
                     assert "upgrade" in result.output.lower() or "portal" in result.output.lower()
@@ -436,7 +436,7 @@ class TestBillingTopupCommand:
                     mock_client.return_value.__enter__.return_value.post.return_value = mock_response
 
                     with patch("webbrowser.open") as mock_browser:
-                        result = runner.invoke(app, ["billing", "topup", "--no-browser"])
+                        result = runner.invoke(app, ["config", "billing", "topup", "--no-browser"])
 
                         assert result.exit_code == 0
                         assert "https://checkout.stripe.com/test" in result.output
@@ -460,7 +460,7 @@ class TestBillingPortalCommand:
                 )
                 mock_client.return_value.__enter__.return_value.post.return_value = mock_response
 
-                result = runner.invoke(app, ["billing", "portal"])
+                result = runner.invoke(app, ["config", "billing", "portal"])
 
                 assert result.exit_code != 0
                 assert "login" in result.output.lower()
@@ -483,7 +483,7 @@ class TestBillingPortalCommand:
                     mock_client.return_value.__enter__.return_value.post.return_value = mock_response
 
                     with patch("webbrowser.open") as mock_browser:
-                        result = runner.invoke(app, ["billing", "portal"])
+                        result = runner.invoke(app, ["config", "billing", "portal"])
 
                         assert result.exit_code == 0
                         mock_browser.assert_called_once_with("https://billing.stripe.com/test")
@@ -506,7 +506,7 @@ class TestBillingPortalCommand:
                     mock_client.return_value.__enter__.return_value.post.return_value = mock_response
 
                     with patch("webbrowser.open") as mock_browser:
-                        result = runner.invoke(app, ["billing", "portal", "--no-browser"])
+                        result = runner.invoke(app, ["config", "billing", "portal", "--no-browser"])
 
                         assert result.exit_code == 0
                         assert "https://billing.stripe.com/test" in result.output
@@ -528,4 +528,4 @@ class TestInsufficientCreditsError:
         message = _friendly_error(402, '{"detail": "Insufficient credits"}', "test-workspace")
 
         assert "credit" in message.lower()
-        assert "wafer billing" in message.lower()
+        assert "wafer config billing" in message.lower()

{wafer_cli-0.2.24 → wafer_cli-0.2.26}/wafer/GUIDE.md

@@ -7,7 +7,7 @@ GPU development primitives for LLM agents.
 Run code on cloud GPUs instantly with workspaces:
 
 ```bash
-wafer login                              # One-time auth
+wafer auth login                         # One-time auth
 wafer workspaces create dev --gpu B200   # Create workspace (NVIDIA B200)
 wafer workspaces exec dev -- python -c "import torch; print(torch.cuda.get_device_name(0))"
 wafer workspaces sync dev ./my-project   # Sync files

wafer_cli-0.2.26/wafer/agent_defaults.py

@@ -0,0 +1,42 @@
+"""Shared agent defaults for kernel optimization tasks.
+
+Single source of truth for bash allowlists and enabled tools used by both:
+- CLI templates (apps/wafer-cli/wafer/templates/optimize_kernelbench.py)
+- Eval configs (research/evals/optimize_kernelbench_eval/.../base_config.py)
+
+Import from here instead of defining your own copy.
+"""
+
+from __future__ import annotations
+
+# Tools available to the agent (coding environment tools)
+ENABLED_TOOLS: list[str] = ["read", "write", "edit", "glob", "grep", "bash"]
+
+# Bash commands allowed for kernel optimization agents.
+# Uses prefix matching — "wafer evaluate" also allows "wafer evaluate kernelbench".
+KERNELBENCH_BASH_ALLOWLIST: list[str] = [
+    # Kernel evaluation
+    "wafer evaluate",
+    # Profiling — AMD
+    "wafer amd rocprof-compute",
+    "wafer amd rocprof-sdk",
+    "wafer amd rocprof-systems",
+    # Profiling — NVIDIA
+    "wafer nvidia ncu",
+    "wafer nvidia nsys",
+    # Analysis
+    "wafer compiler-analyze",
+    # Sub-agents
+    "wafer agent -t ask-docs",
+    # General utilities
+    "python",
+    "python3",
+    "timeout",
+    "ls",
+    "cat",
+    "head",
+    "tail",
+    "wc",
+    "pwd",
+    "which",
+]

{wafer_cli-0.2.24 → wafer_cli-0.2.26}/wafer/billing.py

@@ -1,6 +1,6 @@
 """Billing CLI - Manage credits and subscription.
 
-This module provides the implementation for the `wafer billing` subcommand.
+This module provides the implementation for the `wafer config billing` subcommand.
 """
 
 import json
@@ -126,7 +126,7 @@ def format_usage_text(usage: dict) -> str:
         lines.extend([
             "",
             "Upgrade to Pro for hardware counters and credit topups:",
-            "  wafer billing portal",
+            "  wafer config billing portal",
         ])
 
     return "\n".join(lines)
@@ -153,7 +153,7 @@ def get_usage(json_output: bool = False) -> str:
             usage = response.json()
     except httpx.HTTPStatusError as e:
         if e.response.status_code == 401:
-            raise RuntimeError("Not authenticated. Run: wafer login") from e
+            raise RuntimeError("Not authenticated. Run: wafer auth login") from e
         raise RuntimeError(f"API error: {e.response.status_code} - {e.response.text}") from e
     except httpx.RequestError as e:
         raise RuntimeError(f"Could not reach API: {e}") from e
@@ -188,7 +188,7 @@ def create_topup(amount_cents: int) -> dict:
             return response.json()
     except httpx.HTTPStatusError as e:
         if e.response.status_code == 401:
-            raise RuntimeError("Not authenticated. Run: wafer login") from e
+            raise RuntimeError("Not authenticated. Run: wafer auth login") from e
         if e.response.status_code == 400:
             # Invalid amount
             try:
@@ -200,7 +200,7 @@ def create_topup(amount_cents: int) -> dict:
             # Start tier or other restriction
             raise RuntimeError(
                 "Topup not available for your subscription tier.\n"
-                "Upgrade your subscription first: wafer billing portal"
+                "Upgrade your subscription first: wafer config billing portal"
             ) from e
         if e.response.status_code == 503:
             raise RuntimeError("Billing service temporarily unavailable. Please try again later.") from e
@@ -227,7 +227,7 @@ def get_portal_url() -> dict:
             return response.json()
     except httpx.HTTPStatusError as e:
         if e.response.status_code == 401:
-            raise RuntimeError("Not authenticated. Run: wafer login") from e
+            raise RuntimeError("Not authenticated. Run: wafer auth login") from e
         raise RuntimeError(f"API error: {e.response.status_code} - {e.response.text}") from e
     except httpx.RequestError as e:
         raise RuntimeError(f"Could not reach API: {e}") from e