spice-mcp 0.1.0__tar.gz → 0.1.1__tar.gz

{spice_mcp-0.1.0 → spice_mcp-0.1.1}/.github/workflows/ci.yml

@@ -1,8 +1,8 @@
-name: CI
+name: CI (disabled by default)
 
 on:
-  push:
-  pull_request:
+  # Disable automatic triggers; allow manual runs only
+  workflow_dispatch:
 
 permissions:
   contents: read
@@ -48,7 +48,7 @@ jobs:
       - name: Install dependencies (uv)
         run: |
           uv --version
-          uv sync --frozen --dev
+          uv sync --frozen --group dev
 
       - name: Ruff lint
         run: uv run ruff check --output-format=github --color always .
@@ -93,7 +93,7 @@ jobs:
       - name: Install dependencies (uv)
         run: |
           uv --version
-          uv sync --frozen --dev
+          uv sync --frozen --group dev
 
       - name: mypy
         env:
@@ -132,7 +132,7 @@ jobs:
       - name: Install dependencies (uv)
         run: |
           uv --version
-          uv sync --frozen --dev
+          uv sync --frozen --group dev
 
       - name: Run tests (skip live) with coverage
         env:

{spice_mcp-0.1.0 → spice_mcp-0.1.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: spice-mcp
-Version: 0.1.0
+Version: 0.1.1
 Summary: MCP server for Dune Analytics data access
 Author-email: Evan-Kim2028 <ekcopersonal@gmail.com>
 License-File: LICENSE
@@ -21,27 +21,87 @@ Description-Content-Type: text/markdown
 
 # spice-mcp
 
-spice-mcp is an MCP server for Dune Analytics. It wraps a curated subset of the original Spice client inside a clean architecture (`core` models/ports → `adapters.dune` → service layer → FastMCP tools) and adds agent-friendly workflows for discovery and Sui package exploration. Results are Polars-first in Python and compact, token-efficient in MCP responses.
+spice-mcp is an MCP server for [Dune](https://dune.com/) Analytics. It wraps a curated subset of the original Spice client inside a clean architecture (`core` models/ports → `adapters.dune` → service layer → FastMCP tools) and adds agent-friendly workflows for discovery and Sui package exploration. Results are Polars-first in Python and compact, token-efficient in MCP responses.
 
 Requirements: Python 3.13+
 
 This project uses FastMCP for typed, decorator-registered tools and resources.
 
-Highlights
+## Highlights
 - Polars LazyFrame-first pipeline: results stay lazy until explicitly materialized
-- Ports/adapters layering for maintainable integrations (`docs/architecture.md`)
+- Ports/adapters layering for maintainable integrations ([docs/architecture.md](docs/architecture.md))
 - Discovery utilities (find schemas/tables, describe columns)
 - Sui package workflows (events/transactions/objects) with safe defaults
 - JSONL query history + SQL artifacts (SHA-256) for reproducibility
 - Rich MCP surface: query info/run, discovery, health, Sui, and Dune admin (create/update/fork)
 
-Quick Start
+## What is Dune?
+[Dune](https://dune.com/) is a crypto data platform providing curated blockchain datasets and a public API to run and fetch query results. See the [Dune Docs](https://dune.com/docs) and [Dune API](https://dune.com/docs/api/) for full details.
+
+## Quick Start
 - Export `DUNE_API_KEY` in your shell (the server can also load a local `.env`; set `SPICE_MCP_SKIP_DOTENV=1` to skip during tests).
 - Install dependencies (`uv sync` or `pip install -e .`).
 - Start the FastMCP stdio server:
   - `python -m spice_mcp.mcp.server --env PYTHONPATH=$(pwd)/src`
   - or install the console script via `uv tool install .` and run `spice-mcp`.
 
+## Cursor IDE Setup
+
+To use spice-mcp with Cursor IDE:
+
+1. **Install the MCP Server**:
+   ```bash
+   # Install dependencies and package
+   uv sync
+   pip install -e .
+   
+   # Or install via uv tool (creates console script)
+   uv tool install .
+   ```
+
+2. **Configure Cursor**:
+   - Open Cursor Settings → MCP Servers
+   - Add new MCP server configuration:
+   ```json
+   {
+     "name": "spice-mcp",
+     "command": "spice-mcp",
+     "env": {
+       "DUNE_API_KEY": "your-dune-api-key-here"
+     },
+     "disabled": false
+   }
+   ```
+   Alternatively, if you prefer running from source:
+   ```json
+   {
+     "name": "spice-mcp", 
+     "command": "python",
+     "args": ["-m", "spice_mcp.mcp.server"],
+     "env": {
+       "PYTHONPATH": "/path/to/your/spice-mcp/src",
+       "DUNE_API_KEY": "your-dune-api-key-here"
+     },
+     "disabled": false
+   }
+   ```
+
+3. **Restart Cursor** to load the MCP server
+
+4. **Verify Connection**:
+   - Open Cursor and use the command palette (Cmd/Ctrl + Shift + P)
+   - Search for "MCP" or "spice" commands
+   - Test with `dune_health_check` to verify the connection
+
+5. **Available Tools in Cursor**:
+   - `dune_query`: Run Dune queries by ID, URL, or raw SQL
+   - `dune_find_tables`: Search schemas and list tables
+   - `dune_describe_table`: Get column metadata
+   - `sui_package_overview`: Analyze Sui packages
+   - `dune_health_check`: Verify API connection
+
+**Tip**: Create a `.env` file in your project root with `DUNE_API_KEY=your-key-here` for easier configuration.
+
 ## MCP Tools and Features
 
 All tools expose typed parameters, titles, and tags; failures return a consistent error envelope.
@@ -114,16 +174,16 @@ Core Tools (with parameters)
   - Use: Verify API key presence and logging paths
   - Output: `api_key_present`, `query_history_path`, `logging_enabled`, `status`
 
-Docs
-- See `docs/index.md` for full documentation:
-  - Dune API structure and capabilities: `docs/dune_api.md`
-  - Discovery patterns and examples: `docs/discovery.md`
-  - Sui package workflows: `docs/sui_packages.md`
-  - Tool reference and schemas: `docs/tools.md`
-  - Codex CLI + tooling integration: `docs/codex_cli.md`, `docs/codex_cli_tools.md`
-  - Architecture overview: `docs/architecture.md`
-  - Installation and configuration: `docs/installation.md`, `docs/config.md`
-  - Development and linting: `docs/development.md`
+## Docs
+- See [docs/index.md](docs/index.md) for full documentation:
+  - Dune API structure and capabilities: [docs/dune_api.md](docs/dune_api.md)
+  - Discovery patterns and examples: [docs/discovery.md](docs/discovery.md)
+  - Sui package workflows: [docs/sui_packages.md](docs/sui_packages.md)
+  - Tool reference and schemas: [docs/tools.md](docs/tools.md)
+  - Codex CLI + tooling integration: [docs/codex_cli.md](docs/codex_cli.md), [docs/codex_cli_tools.md](docs/codex_cli_tools.md)
+  - Architecture overview: [docs/architecture.md](docs/architecture.md)
+  - Installation and configuration: [docs/installation.md](docs/installation.md), [docs/config.md](docs/config.md)
+  - Development and linting: [docs/development.md](docs/development.md)
 
 Notes
 - Legacy Spice code now lives under `src/spice_mcp/adapters/dune` (extract, cache, urls, types).

{spice_mcp-0.1.0 → spice_mcp-0.1.1}/README.md

@@ -1,26 +1,86 @@
 # spice-mcp
 
-spice-mcp is an MCP server for Dune Analytics. It wraps a curated subset of the original Spice client inside a clean architecture (`core` models/ports → `adapters.dune` → service layer → FastMCP tools) and adds agent-friendly workflows for discovery and Sui package exploration. Results are Polars-first in Python and compact, token-efficient in MCP responses.
+spice-mcp is an MCP server for [Dune](https://dune.com/) Analytics. It wraps a curated subset of the original Spice client inside a clean architecture (`core` models/ports → `adapters.dune` → service layer → FastMCP tools) and adds agent-friendly workflows for discovery and Sui package exploration. Results are Polars-first in Python and compact, token-efficient in MCP responses.
 
 Requirements: Python 3.13+
 
 This project uses FastMCP for typed, decorator-registered tools and resources.
 
-Highlights
+## Highlights
 - Polars LazyFrame-first pipeline: results stay lazy until explicitly materialized
-- Ports/adapters layering for maintainable integrations (`docs/architecture.md`)
+- Ports/adapters layering for maintainable integrations ([docs/architecture.md](docs/architecture.md))
 - Discovery utilities (find schemas/tables, describe columns)
 - Sui package workflows (events/transactions/objects) with safe defaults
 - JSONL query history + SQL artifacts (SHA-256) for reproducibility
 - Rich MCP surface: query info/run, discovery, health, Sui, and Dune admin (create/update/fork)
 
-Quick Start
+## What is Dune?
+[Dune](https://dune.com/) is a crypto data platform providing curated blockchain datasets and a public API to run and fetch query results. See the [Dune Docs](https://dune.com/docs) and [Dune API](https://dune.com/docs/api/) for full details.
+
+## Quick Start
 - Export `DUNE_API_KEY` in your shell (the server can also load a local `.env`; set `SPICE_MCP_SKIP_DOTENV=1` to skip during tests).
 - Install dependencies (`uv sync` or `pip install -e .`).
 - Start the FastMCP stdio server:
   - `python -m spice_mcp.mcp.server --env PYTHONPATH=$(pwd)/src`
   - or install the console script via `uv tool install .` and run `spice-mcp`.
 
+## Cursor IDE Setup
+
+To use spice-mcp with Cursor IDE:
+
+1. **Install the MCP Server**:
+   ```bash
+   # Install dependencies and package
+   uv sync
+   pip install -e .
+   
+   # Or install via uv tool (creates console script)
+   uv tool install .
+   ```
+
+2. **Configure Cursor**:
+   - Open Cursor Settings → MCP Servers
+   - Add new MCP server configuration:
+   ```json
+   {
+     "name": "spice-mcp",
+     "command": "spice-mcp",
+     "env": {
+       "DUNE_API_KEY": "your-dune-api-key-here"
+     },
+     "disabled": false
+   }
+   ```
+   Alternatively, if you prefer running from source:
+   ```json
+   {
+     "name": "spice-mcp", 
+     "command": "python",
+     "args": ["-m", "spice_mcp.mcp.server"],
+     "env": {
+       "PYTHONPATH": "/path/to/your/spice-mcp/src",
+       "DUNE_API_KEY": "your-dune-api-key-here"
+     },
+     "disabled": false
+   }
+   ```
+
+3. **Restart Cursor** to load the MCP server
+
+4. **Verify Connection**:
+   - Open Cursor and use the command palette (Cmd/Ctrl + Shift + P)
+   - Search for "MCP" or "spice" commands
+   - Test with `dune_health_check` to verify the connection
+
+5. **Available Tools in Cursor**:
+   - `dune_query`: Run Dune queries by ID, URL, or raw SQL
+   - `dune_find_tables`: Search schemas and list tables
+   - `dune_describe_table`: Get column metadata
+   - `sui_package_overview`: Analyze Sui packages
+   - `dune_health_check`: Verify API connection
+
+**Tip**: Create a `.env` file in your project root with `DUNE_API_KEY=your-key-here` for easier configuration.
+
 ## MCP Tools and Features
 
 All tools expose typed parameters, titles, and tags; failures return a consistent error envelope.
@@ -93,16 +153,16 @@ Core Tools (with parameters)
   - Use: Verify API key presence and logging paths
   - Output: `api_key_present`, `query_history_path`, `logging_enabled`, `status`
 
-Docs
-- See `docs/index.md` for full documentation:
-  - Dune API structure and capabilities: `docs/dune_api.md`
-  - Discovery patterns and examples: `docs/discovery.md`
-  - Sui package workflows: `docs/sui_packages.md`
-  - Tool reference and schemas: `docs/tools.md`
-  - Codex CLI + tooling integration: `docs/codex_cli.md`, `docs/codex_cli_tools.md`
-  - Architecture overview: `docs/architecture.md`
-  - Installation and configuration: `docs/installation.md`, `docs/config.md`
-  - Development and linting: `docs/development.md`
+## Docs
+- See [docs/index.md](docs/index.md) for full documentation:
+  - Dune API structure and capabilities: [docs/dune_api.md](docs/dune_api.md)
+  - Discovery patterns and examples: [docs/discovery.md](docs/discovery.md)
+  - Sui package workflows: [docs/sui_packages.md](docs/sui_packages.md)
+  - Tool reference and schemas: [docs/tools.md](docs/tools.md)
+  - Codex CLI + tooling integration: [docs/codex_cli.md](docs/codex_cli.md), [docs/codex_cli_tools.md](docs/codex_cli_tools.md)
+  - Architecture overview: [docs/architecture.md](docs/architecture.md)
+  - Installation and configuration: [docs/installation.md](docs/installation.md), [docs/config.md](docs/config.md)
+  - Development and linting: [docs/development.md](docs/development.md)
 
 Notes
 - Legacy Spice code now lives under `src/spice_mcp/adapters/dune` (extract, cache, urls, types).

{spice_mcp-0.1.0 → spice_mcp-0.1.1}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "spice-mcp"
-version = "0.1.0"
+version = "0.1.1"
 description = "MCP server for Dune Analytics data access"
 authors = [
     { name = "Evan-Kim2028", email = "ekcopersonal@gmail.com" }
@@ -36,6 +36,23 @@ managed = true
 dev-dependencies = [
     "pytest>=8.0.0",
     "pytest-asyncio>=0.23.0",
+    "pytest-cov>=5.0.0",
+    "freezegun>=1.4.0",
+    "jsonschema>=4.21.1",
+    "responses>=0.25.0",
+    "aioresponses>=0.7.6",
+    "vcrpy>=6.0.1",
+    "hypothesis>=6.100.0",
+    "ruff>=0.6.9",
+    "mypy>=1.11.2",
+]
+
+# uv dependency groups for CI and local dev
+[dependency-groups]
+dev = [
+    "pytest>=8.0.0",
+    "pytest-asyncio>=0.23.0",
+    "pytest-cov>=5.0.0",
     "freezegun>=1.4.0",
     "jsonschema>=4.21.1",
     "responses>=0.25.0",

spice_mcp-0.1.1/scripts/bridgez/make_circle_comparison.py

@@ -0,0 +1,114 @@
+import csv
+from pathlib import Path
+
+
+BASE = Path("notes/bridgez")
+CIRCLE_DIR = BASE / "circle"
+
+
+def read_single_row_metrics(path: Path):
+    with path.open() as f:
+        reader = csv.DictReader(f)
+        row = next(reader)
+        return {
+            "tx_count": float(row.get("tx_count", 0) or 0),
+            "total_usd": float(row.get("total_usd", 0) or 0),
+            "avg_usd": float(row.get("avg_usd", 0) or 0),
+        }
+
+
+def read_circle_totals(path: Path):
+    rows = []
+    with path.open() as f:
+        reader = csv.DictReader(f)
+        for r in reader:
+            rows.append(
+                {
+                    "schema": r["schema"],
+                    "usd_out_30d": float(r["usd_out_30d"]) if r["usd_out_30d"] else 0.0,
+                    "usd_out_ytd": float(r["usd_out_ytd"]) if r["usd_out_ytd"] else 0.0,
+                }
+            )
+    return rows
+
+
+def write_text(path: Path, content: str):
+    path.write_text(content)
+
+
+def update_readme(readme_path: Path, section: str):
+    # Append or replace a Circle vs LayerZero section
+    marker_start = "\n## Circle vs LayerZero (Summary)\n"
+    content = readme_path.read_text() if readme_path.exists() else ""
+    if "## Circle vs LayerZero (Summary)" in content:
+        # replace from marker to end or next header
+        parts = content.split("## Circle vs LayerZero (Summary)")
+        head = parts[0]
+        tail = "## Circle vs LayerZero (Summary)" + parts[1]
+        # cut tail at the next header if present
+        tail_lines = tail.split("\n## ")
+        new_tail = "## Circle vs LayerZero (Summary)\n" + section
+        if len(tail_lines) > 1:
+            # keep next headers
+            new_content = head + new_tail + "\n## " + "\n## ".join(tail_lines[1:])
+        else:
+            new_content = head + new_tail
+        readme_path.write_text(new_content)
+    else:
+        readme_path.write_text(content.rstrip() + marker_start + section)
+
+
+def fmt_usd(x: float) -> str:
+    return f"${x:,.2f}"
+
+
+def main():
+    # LayerZero metrics
+    lz_30d = read_single_row_metrics(BASE / "metrics_last_30d.csv")
+    lz_ytd = read_single_row_metrics(BASE / "metrics_ytd.csv")
+
+    # Circle per-schema summary
+    circle_rows = read_circle_totals(CIRCLE_DIR / "flows_outbound_summary.csv")
+    c_30d_total = sum(r["usd_out_30d"] for r in circle_rows)
+    c_ytd_total = sum(r["usd_out_ytd"] for r in circle_rows)
+
+    # Build comparison text
+    pct_30d = (c_30d_total / lz_30d["total_usd"] * 100.0) if lz_30d["total_usd"] else 0.0
+    pct_ytd = (c_ytd_total / lz_ytd["total_usd"] * 100.0) if lz_ytd["total_usd"] else 0.0
+
+    top_circle_30d = sorted(circle_rows, key=lambda r: r["usd_out_30d"], reverse=True)[:5]
+    top_circle_ytd = sorted(circle_rows, key=lambda r: r["usd_out_ytd"], reverse=True)[:5]
+
+    lines = []
+    lines.append("Circle vs LayerZero (USD volumes)\n")
+    lines.append(f"- LayerZero — 30d: {fmt_usd(lz_30d['total_usd'])}; YTD: {fmt_usd(lz_ytd['total_usd'])}")
+    lines.append(f"- Circle (CCTP outbound) — 30d: {fmt_usd(c_30d_total)} ({pct_30d:.1f}% of LZ); YTD: {fmt_usd(c_ytd_total)} ({pct_ytd:.1f}% of LZ)\n")
+    lines.append("Top Circle schemas (30d):")
+    for r in top_circle_30d:
+        lines.append(f"- {r['schema']}: {fmt_usd(r['usd_out_30d'])}")
+    lines.append("")
+    lines.append("Top Circle schemas (YTD):")
+    for r in top_circle_ytd:
+        lines.append(f"- {r['schema']}: {fmt_usd(r['usd_out_ytd'])}")
+    lines.append("")
+    lines.append("Note: circle_bnb and circle_zksync may be missing if their tables/columns differ; totals reflect included schemas only.")
+
+    text = "\n".join(lines) + "\n"
+
+    # Write dedicated comparison under circle
+    write_text(CIRCLE_DIR / "COMPARISON.txt", text)
+
+    # Update top-level README with a concise section
+    section = (
+        f"\n- LayerZero — 30d: {fmt_usd(lz_30d['total_usd'])}; YTD: {fmt_usd(lz_ytd['total_usd'])}\n"
+        f"- Circle (CCTP outbound) — 30d: {fmt_usd(c_30d_total)} ({pct_30d:.1f}% of LZ); YTD: {fmt_usd(c_ytd_total)} ({pct_ytd:.1f}% of LZ)\n"
+        "- Top Circle schemas (30d): "
+        + ", ".join([f"{r['schema']} {fmt_usd(r['usd_out_30d'])}" for r in top_circle_30d])
+        + "\n"
+    )
+    update_readme(BASE / "README.md", section)
+
+
+if __name__ == "__main__":
+    main()
+

{spice_mcp-0.1.0 → spice_mcp-0.1.1}/src/spice_mcp/adapters/dune/extract.py

@@ -3,6 +3,7 @@ from __future__ import annotations
 import io
 import time
 from typing import TYPE_CHECKING, overload
+import os
 
 from ..http_client import HttpClient
 from . import cache as _cache
@@ -32,6 +33,10 @@ from .typing_utils import resolve_raw_sql_template_id
 
 ADAPTER_VERSION = "spice-mcp-adapter"
 
+# Runtime-configurable HTTP timeouts (helps avoid host-level timeouts)
+_GET_TIMEOUT: float = float(os.getenv("SPICE_DUNE_GET_TIMEOUT", os.getenv("SPICE_HTTP_TIMEOUT", "30")))
+_POST_TIMEOUT: float = float(os.getenv("SPICE_DUNE_POST_TIMEOUT", os.getenv("SPICE_HTTP_TIMEOUT", "30")))
+
 if TYPE_CHECKING:
     from collections.abc import Mapping, Sequence
     from typing import Any, Literal
@@ -749,7 +754,7 @@ def _execute(
         print('executing query, query_id = ' + str(query_id))
 
     # perform request
-    response = _transport_post(url, headers=headers, json=data, timeout=15.0)
+    response = _transport_post(url, headers=headers, json=data, timeout=_POST_TIMEOUT)
     result: Mapping[str, Any] = response.json()
 
     # check for errors
@@ -786,7 +791,7 @@ async def _async_execute(
         print('executing query, query_id = ' + str(query_id))
 
     # perform request
-    timeout = aiohttp.ClientTimeout(total=30)
+    timeout = aiohttp.ClientTimeout(total=_POST_TIMEOUT)
     async with aiohttp.ClientSession(timeout=timeout) as session:
         async with session.post(url, headers=headers, json=data) as response:
             result: Mapping[str, Any] = await response.json()
@@ -862,12 +867,12 @@ def _get_results(
     def _get_with_retries(u: str):
         client = current_http_client()
         if client is not None:
-            return client.request("GET", u, headers=headers, timeout=30.0)
+            return client.request("GET", u, headers=headers, timeout=_GET_TIMEOUT)
 
         attempts = 0
         backoff = 0.5
         while True:
-            resp = _transport_get(u, headers=headers, timeout=30.0)
+            resp = _transport_get(u, headers=headers, timeout=_GET_TIMEOUT)
             if resp.status_code in (429, 502):
                 attempts += 1
                 if attempts >= 3:
@@ -963,7 +968,7 @@ async def _async_get_results(
             print('getting results, execution_id = ' + str(execution['execution_id']))
 
     # perform request
-    timeout = aiohttp.ClientTimeout(total=30)
+    timeout = aiohttp.ClientTimeout(total=_GET_TIMEOUT)
     async with aiohttp.ClientSession(timeout=timeout) as session:
         # GET with simple retry/backoff for 429/502
         attempts = 0
@@ -1169,7 +1174,7 @@ def _poll_execution(
             )
 
         # poll
-        response = _http_get(url, headers=headers, timeout=15.0)
+        response = _http_get(url, headers=headers, timeout=_GET_TIMEOUT)
         result = response.json()
         if (
             'is_execution_finished' not in result
@@ -1243,7 +1248,7 @@ async def _async_poll_execution(
     t_start = time.time()
 
     # poll until completion
-    timeout = aiohttp.ClientTimeout(total=30)
+    timeout = aiohttp.ClientTimeout(total=_GET_TIMEOUT)
     async with aiohttp.ClientSession(timeout=timeout) as session:
         sleep_amount = poll_interval
         while True:

{spice_mcp-0.1.0 → spice_mcp-0.1.1}/src/spice_mcp/adapters/dune/typing_utils.py

@@ -1,5 +1,7 @@
 from __future__ import annotations
 
+import os
+
 
 def resolve_raw_sql_template_id() -> int:
     """Return a stable template ID used for executing raw SQL text.
@@ -7,4 +9,4 @@ def resolve_raw_sql_template_id() -> int:
     Tests stub HTTP boundaries and only require a consistent integer. This
     placeholder can be adjusted if upstream semantics change.
     """
-    return 0
+    return int(os.getenv("SPICE_RAW_SQL_QUERY_ID", "4060379"))

spice_mcp-0.1.1/tests/offline/test_typing_utils.py

@@ -0,0 +1,62 @@
+"""Tests for typing utilities, especially environment variable handling."""
+
+from __future__ import annotations
+
+import os
+from unittest.mock import patch
+
+from spice_mcp.adapters.dune.typing_utils import resolve_raw_sql_template_id
+
+
+def test_resolve_raw_sql_template_id_default():
+    """Test that the function returns the default template ID when no env var is set."""
+    # Ensure env var is not set
+    with patch.dict(os.environ, {}, clear=True):
+        result = resolve_raw_sql_template_id()
+        assert result == 4060379
+        assert isinstance(result, int)
+
+
+def test_resolve_raw_sql_template_id_from_env():
+    """Test that the function reads from the SPICE_RAW_SQL_QUERY_ID environment variable."""
+    custom_id = "12345"
+    with patch.dict(os.environ, {"SPICE_RAW_SQL_QUERY_ID": custom_id}):
+        result = resolve_raw_sql_template_id()
+        assert result == int(custom_id)
+
+
+def test_resolve_raw_sql_template_id_invalid_env():
+    """Test that the function handles invalid environment variable values gracefully."""
+    with patch.dict(os.environ, {"SPICE_RAW_SQL_QUERY_ID": "invalid_number"}):
+        # This should raise a ValueError when trying to convert to int
+        try:
+            resolve_raw_sql_template_id()
+            assert False, "Expected ValueError for invalid number"
+        except ValueError:
+            pass  # Expected
+
+
+def test_resolve_raw_sql_template_id_zero_env():
+    """Test that the function handles zero value correctly."""
+    with patch.dict(os.environ, {"SPICE_RAW_SQL_QUERY_ID": "0"}):
+        result = resolve_raw_sql_template_id()
+        assert result == 0
+
+
+def test_determine_input_type_raw_sql_uses_template_id(monkeypatch):
+    """Test that determine_input_type correctly resolves raw SQL using the template ID."""
+    from spice_mcp.adapters.dune.extract import determine_input_type
+    
+    # Test with default template ID
+    with patch.dict(os.environ, {}, clear=True):
+        query_id, execution, parameters = determine_input_type("SELECT 1 as test")
+        assert query_id == 4060379
+        assert execution is None
+        assert parameters == {"query": "SELECT 1 as test"}
+    
+    # Test with custom template ID
+    with patch.dict(os.environ, {"SPICE_RAW_SQL_QUERY_ID": "12345"}):
+        query_id, execution, parameters = determine_input_type("SELECT count(*) as total")
+        assert query_id == 12345
+        assert execution is None
+        assert parameters == {"query": "SELECT count(*) as total"}