Add 4 tools for getting grafana dashboards #1075
Conversation
Uses this config
toolsets:
grafana/dashboards:
enabled: true
config:
url: http://localhost:8118
healthcheck: "healthz"
api_key: <Service account token>
WalkthroughAdds a Grafana dashboards toolset and config, a BaseGrafanaTool with a unified request pipeline, four dashboard tools (SearchDashboards, GetDashboardByUID, GetHomeDashboard, GetDashboardTags), a Jinja2 toolset template, documentation/navigation, a small docstring fix, and a Toolset helper to load LLM instruction files from disk. Changes
Sequence Diagram(s)sequenceDiagram
participant Agent as Agent/User
participant Tool as GrafanaTool (Search/Get/etc)
participant Base as BaseGrafanaTool._make_grafana_request
participant Grafana as Grafana API
Agent->>Tool: invoke(params, context)
Note right of Tool #eef7ff: compose endpoint & query_params
Tool->>Base: _make_grafana_request(endpoint, params, query_params)
Base->>Grafana: HTTP GET (Authorization header, built URL)
Grafana-->>Base: response / error
Base-->>Tool: StructuredToolResult
Tool-->>Agent: StructuredToolResult
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes Points to review closely:
Suggested reviewers
Pre-merge checks and finishing touches❌ Failed checks (1 warning, 1 inconclusive)
✅ Passed checks (1 passed)
✨ Finishing touches
🧪 Generate unit tests (beta)
📜 Recent review detailsConfiguration used: CodeRabbit UI Review profile: CHILL Plan: Pro 📒 Files selected for processing (1)
🧰 Additional context used📓 Path-based instructions (2)**/*.py📄 CodeRabbit inference engine (CLAUDE.md)
Files:
holmes/plugins/toolsets/**/*📄 CodeRabbit inference engine (CLAUDE.md)
Files:
🧬 Code graph analysis (1)holmes/plugins/toolsets/grafana/toolset_grafana.py (4)
🪛 Ruff (0.14.1)holmes/plugins/toolsets/grafana/toolset_grafana.py154-154: Unused method argument: (ARG002) 191-191: Unused method argument: (ARG002) 210-210: Unused method argument: (ARG002) 227-227: Unused method argument: (ARG002) 230-230: Unused method argument: (ARG002) 243-243: Unused method argument: (ARG002) 246-246: Unused method argument: (ARG002) ⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (3)
🔇 Additional comments (2)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (5)
holmes/plugins/toolsets/grafana/toolset_grafana.py (5)
107-141: Simplify query param building; accept both arrays and CSV; pass booleans and clamp limit.Support both array inputs (preferred) and legacy comma-separated strings. Keep booleans as bool and clamp limit to API cap.
- query_params = {} - if params.get("query"): - query_params["query"] = params["query"] - if params.get("tag"): - query_params["tag"] = params["tag"] - if params.get("type"): - query_params["type"] = params["type"] - if params.get("dashboardIds"): - # Check if dashboardIds also needs to be passed as multiple params - dashboard_ids = params["dashboardIds"].split(",") - query_params["dashboardIds"] = [ - dashboard_id.strip() - for dashboard_id in dashboard_ids - if dashboard_id.strip() - ] - if params.get("dashboardUIDs"): - # Handle dashboardUIDs as a list - split comma-separated values - dashboard_uids = params["dashboardUIDs"].split(",") - query_params["dashboardUIDs"] = [ - uid.strip() for uid in dashboard_uids if uid.strip() - ] - if params.get("folderUIDs"): - # Check if folderUIDs also needs to be passed as multiple params - folder_uids = params["folderUIDs"].split(",") - query_params["folderUIDs"] = [ - uid.strip() for uid in folder_uids if uid.strip() - ] - if params.get("starred") is not None: - query_params["starred"] = str(params["starred"]).lower() - if params.get("limit"): - query_params["limit"] = params["limit"] - if params.get("page"): - query_params["page"] = params["page"] + query_params: Dict = {} + for key in ("query", "tag", "type"): + if params.get(key) is not None: + query_params[key] = params[key] + + def to_list(v): + if v is None: + return None + if isinstance(v, list): + return [x for x in v if str(x).strip()] + if isinstance(v, str): + return [x.strip() for x in v.split(",") if x.strip()] + return None + + for key in ("dashboardIds", "dashboardUIDs", "folderUIDs"): + seq = to_list(params.get(key)) + if seq: + query_params[key] = seq + + if params.get("starred") is not None: + query_params["starred"] = bool(params["starred"]) + + if params.get("limit") is not None: + try: + limit = int(params["limit"]) + query_params["limit"] = max(1, min(5000, limit)) + except (TypeError, ValueError): + pass + if params.get("page") is not None: + try: + query_params["page"] = int(params["page"]) + except (TypeError, ValueError): + pass
107-107: Silence Ruff ARG002 by renaming unused params to “_”.Keeps signatures intact while removing linter noise.
-def _invoke(self, params: dict, context: ToolInvokeContext) -> StructuredToolResult: +def _invoke(self, params: dict, _: ToolInvokeContext) -> StructuredToolResult: @@ -def get_parameterized_one_liner(self, params: Dict) -> str: +def get_parameterized_one_liner(self, _: Dict) -> str: @@ -def _invoke(self, params: dict, context: ToolInvokeContext) -> StructuredToolResult: +def _invoke(self, params: dict, _: ToolInvokeContext) -> StructuredToolResult: @@ -def _invoke(self, params: dict, context: ToolInvokeContext) -> StructuredToolResult: +def _invoke(self, params: dict, _: ToolInvokeContext) -> StructuredToolResult: @@ -def get_parameterized_one_liner(self, params: Dict) -> str: +def get_parameterized_one_liner(self, _: Dict) -> str: @@ -def _invoke(self, params: dict, context: ToolInvokeContext) -> StructuredToolResult: +def _invoke(self, params: dict, _: ToolInvokeContext) -> StructuredToolResult: @@ -def get_parameterized_one_liner(self, params: Dict) -> str: +def get_parameterized_one_liner(self, _: Dict) -> str:As per coding guidelines (Ruff).
Also applies to: 144-144, 163-163, 180-180, 183-183, 196-196, 199-199
12-12: Remove unnecessary “type: ignore” on requests import.
requestsis typed; prefer clean import.-import requests # type: ignore +import requestsAs per typing/mypy hygiene.
206-215: Set docs_url to Dashboard HTTP API and replace icon with official Grafana asset.The toolset provides dashboard operations (search, get by UID, home, tags). Set docs_url to
https://grafana.com/docs/grafana/latest/developers/http_api/dashboard/for primary dashboard API reference. Replace the third-party CDN icon with an official Grafana-hosted asset (e.g.,https://grafana.com/media/docs/grafana-icon.pngor equivalent from Grafana's official assets).
59-103: Use native array types and enum constraints for improved API alignment and UX.The Grafana HTTP API expects arrays for
dashboardUIDsandfolderUIDsacross v9–v11 and accepts repeated query parameters. Update parameter definitions to reflect this and add enum constraint fortype:"type": ToolParameter( description="Filter by type: 'dash-folder' or 'dash-db'", type="string", + enum=["dash-folder", "dash-db"], required=False, ), "dashboardIds": ToolParameter( - description="List of dashboard IDs to filter (comma-separated)", - type="string", + description="Dashboard numeric IDs to filter", + type="array", + items=ToolParameter(type="integer"), required=False, ), "dashboardUIDs": ToolParameter( - description="List of dashboard UIDs to search for (comma-separated)", - type="string", + description="Dashboard UIDs to search for", + type="array", + items=ToolParameter(type="string"), required=False, ), "folderUIDs": ToolParameter( - description="List of folder UIDs to search within (comma-separated)", - type="string", + description="Folder UIDs to search within", + type="array", + items=ToolParameter(type="string"), required=False, ),The
ToolParameterclass supports bothenumanditemsattributes. Update the_invoke()method to handle actual array inputs directly instead of parsing comma-separated strings.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
holmes/plugins/toolsets/grafana/toolset_grafana.py(2 hunks)
🧰 Additional context used
📓 Path-based instructions (2)
**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
**/*.py: Use Ruff for formatting and linting (configured in pyproject.toml) for all Python code
Type hints are required; code should pass mypy (configured in pyproject.toml)
ALWAYS place Python imports at the top of the file, not inside functions or methods
Files:
holmes/plugins/toolsets/grafana/toolset_grafana.py
holmes/plugins/toolsets/**/*
📄 CodeRabbit inference engine (CLAUDE.md)
Toolsets must be located as holmes/plugins/toolsets/{name}.yaml or holmes/plugins/toolsets/{name}/
Files:
holmes/plugins/toolsets/grafana/toolset_grafana.py
🧬 Code graph analysis (1)
holmes/plugins/toolsets/grafana/toolset_grafana.py (3)
holmes/core/tools.py (5)
Tool(173-365)StructuredToolResult(79-103)StructuredToolResultStatus(52-76)ToolParameter(155-161)ToolInvokeContext(164-170)holmes/plugins/toolsets/grafana/base_grafana_toolset.py (1)
BaseGrafanaToolset(11-54)holmes/plugins/toolsets/utils.py (1)
toolset_name_for_one_liner(232-236)
🪛 Ruff (0.14.1)
holmes/plugins/toolsets/grafana/toolset_grafana.py
40-40: Probable use of requests call without timeout
(S113)
107-107: Unused method argument: context
(ARG002)
144-144: Unused method argument: params
(ARG002)
163-163: Unused method argument: context
(ARG002)
180-180: Unused method argument: context
(ARG002)
183-183: Unused method argument: params
(ARG002)
196-196: Unused method argument: context
(ARG002)
199-199: Unused method argument: params
(ARG002)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (3)
- GitHub Check: llm_evals
- GitHub Check: Pre-commit checks
- GitHub Check: build
🔇 Additional comments (1)
holmes/plugins/toolsets/grafana/toolset_grafana.py (1)
17-23: Good abstraction: shared HTTP plumbing in BaseGrafanaTool.Centralizing request logic reduces duplication and standardizes results. LGTM.
There was a problem hiding this comment.
Actionable comments posted: 0
♻️ Duplicate comments (1)
holmes/plugins/toolsets/grafana/toolset_grafana.py (1)
53-78: Critical: Still missing timeout, error handling, and has URL join issue.This method has the same critical issues previously flagged:
- Line 69: No timeout risks indefinite hangs on request threads.
- Line 66: Using
urljoinwith endpoints that have leading "/" can drop subpaths from the base URL.- Lines 69-71: No exception handling for HTTP errors, connection failures, or JSON decode errors will cause the tool to crash.
As per the previous review, apply error handling with timeout, fix URL joining to preserve subpaths, and catch HTTP/connection/JSON exceptions. The suggested diff in the past review comment provides a complete solution.
🧹 Nitpick comments (1)
holmes/plugins/toolsets/grafana/toolset_grafana.py (1)
192-194: Consider URL-encoding the uid parameter.While Grafana UIDs are typically alphanumeric, the
uidparameter is directly interpolated into the URL path without encoding. If a UID contains special characters, this could cause issues.Apply this diff to add defensive URL encoding:
+from urllib.parse import quotedef _invoke(self, params: dict, context: ToolInvokeContext) -> StructuredToolResult: uid = params["uid"] - return self._make_grafana_request(f"/api/dashboards/uid/{uid}", params) + return self._make_grafana_request(f"/api/dashboards/uid/{quote(uid, safe='')}", params)
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
holmes/plugins/toolsets/grafana/toolset_grafana.py(2 hunks)
🧰 Additional context used
📓 Path-based instructions (2)
**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
**/*.py: Use Ruff for formatting and linting (configured in pyproject.toml) for all Python code
Type hints are required; code should pass mypy (configured in pyproject.toml)
ALWAYS place Python imports at the top of the file, not inside functions or methods
Files:
holmes/plugins/toolsets/grafana/toolset_grafana.py
holmes/plugins/toolsets/**/*
📄 CodeRabbit inference engine (CLAUDE.md)
Toolsets must be located as holmes/plugins/toolsets/{name}.yaml or holmes/plugins/toolsets/{name}/
Files:
holmes/plugins/toolsets/grafana/toolset_grafana.py
🧬 Code graph analysis (1)
holmes/plugins/toolsets/grafana/toolset_grafana.py (4)
holmes/plugins/toolsets/grafana/common.py (2)
get_base_url(52-56)GrafanaConfig(9-22)holmes/plugins/toolsets/utils.py (1)
toolset_name_for_one_liner(232-236)holmes/plugins/toolsets/grafana/base_grafana_toolset.py (1)
BaseGrafanaToolset(11-54)holmes/core/tools.py (5)
Tool(173-365)StructuredToolResult(79-103)StructuredToolResultStatus(52-76)ToolParameter(155-161)ToolInvokeContext(164-170)
🪛 Ruff (0.14.1)
holmes/plugins/toolsets/grafana/toolset_grafana.py
69-69: Probable use of requests call without timeout
(S113)
136-136: Unused method argument: context
(ARG002)
173-173: Unused method argument: params
(ARG002)
192-192: Unused method argument: context
(ARG002)
209-209: Unused method argument: context
(ARG002)
212-212: Unused method argument: params
(ARG002)
225-225: Unused method argument: context
(ARG002)
228-228: Unused method argument: params
(ARG002)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (3)
- GitHub Check: llm_evals
- GitHub Check: Pre-commit checks
- GitHub Check: build
🔇 Additional comments (4)
holmes/plugins/toolsets/grafana/toolset_grafana.py (4)
18-21: LGTM!The configuration class correctly extends
GrafanaConfigand provides a sensible default healthcheck endpoint for the dashboard toolset.
24-43: LGTM!The toolset class is well-structured with proper config_class declaration, tool initialization, and a type-safe property for accessing the configuration.
81-174: LGTM!The search implementation correctly builds query parameters from various optional filters, properly handles comma-separated list inputs, and follows the expected tool interface.
Note: The Ruff warnings about unused
contextandparamsparameters are false positives—these are required by the abstractToolinterface.
200-229: LGTM!Both
GetHomeDashboardandGetDashboardTagsare clean, straightforward implementations that correctly use the base class's request helper.Note: The Ruff warnings about unused
contextandparamsparameters are false positives—these are required by the abstractToolinterface.
There was a problem hiding this comment.
Actionable comments posted: 0
♻️ Duplicate comments (1)
holmes/plugins/toolsets/grafana/toolset_grafana.py (1)
58-83: Critical: Past review issues remain unaddressed—missing timeout, error handling, and safe URL joining.The verification confirms all three issues are still present in the code:
No timeout: When endpoint starts with "/" urljoin replaces the entire base path. If base URL is
http://host/grafana/and endpoint is/api/search, the result becomeshttp://host/api/searchinstead ofhttp://host/grafana/api/search, losing the configured subpath.No timeout on requests.get (Ruff S113): The request can hang indefinitely on slow/unresponsive Grafana instances, blocking the request thread.
No error handling: HTTP errors, connection failures, and JSON decode errors will crash the tool instead of returning a structured error result.
Apply the robust implementation from the previous review:
- def _make_grafana_request( - self, endpoint: str, params: dict, query_params: Optional[Dict] = None - ) -> StructuredToolResult: - """Make a GET request to Grafana API and return structured result. - - Args: - endpoint: API endpoint path (e.g., "/api/search") - params: Original parameters passed to the tool - query_params: Optional query parameters for the request - - Returns: - StructuredToolResult with the API response data - """ - url = urljoin(get_base_url(self._toolset.grafana_config), endpoint) - headers = {"Authorization": f"Bearer {self._toolset.grafana_config.api_key}"} - - response = requests.get(url, headers=headers, params=query_params) - response.raise_for_status() - data = response.json() - - return StructuredToolResult( - status=StructuredToolResultStatus.SUCCESS, - data=data, - url=url, - params=params, - ) + def _make_grafana_request( + self, endpoint: str, params: dict, query_params: Optional[Dict] = None + ) -> StructuredToolResult: + """Make a GET request to Grafana API and return structured result. + + Args: + endpoint: API endpoint path (e.g., "/api/search") + params: Original parameters passed to the tool + query_params: Optional query parameters for the request + + Returns: + StructuredToolResult with the API response data + """ + from json import JSONDecodeError + from requests import exceptions as req_exc + + # Safe URL joining that preserves subpaths + base_url = get_base_url(self._toolset.grafana_config).rstrip("/") + "/" + url = urljoin(base_url, endpoint.lstrip("/")) + + headers = { + "Authorization": f"Bearer {self._toolset.grafana_config.api_key}", + "Accept": "application/json", + } + + # Connect timeout: 3.05s, Read timeout: 10s + timeout = (3.05, 10) + + try: + response = requests.get( + url, + headers=headers, + params=query_params, + timeout=timeout, + ) + response.raise_for_status() + + try: + data = response.json() + except JSONDecodeError: + data = response.text + + return StructuredToolResult( + status=StructuredToolResultStatus.SUCCESS, + data=data, + url=url, + params=params, + return_code=response.status_code, + ) + except req_exc.Timeout as e: + return StructuredToolResult( + status=StructuredToolResultStatus.ERROR, + error=f"Request to Grafana timed out: {e}", + url=url, + params=params, + ) + except req_exc.HTTPError as e: + code = e.response.status_code if hasattr(e, "response") and e.response else None + message = e.response.text if hasattr(e, "response") and e.response else str(e) + return StructuredToolResult( + status=StructuredToolResultStatus.ERROR, + error=message, + return_code=code, + url=url, + params=params, + ) + except req_exc.RequestException as e: + return StructuredToolResult( + status=StructuredToolResultStatus.ERROR, + error=str(e), + url=url, + params=params, + )
🧹 Nitpick comments (1)
holmes/plugins/toolsets/grafana/toolset_grafana.py (1)
141-177: Consider consolidating the repetitive query parameter building logic.The query parameter construction follows a repetitive pattern across multiple parameters. While functional, consolidating this logic could improve maintainability.
Consider extracting a helper method:
def _build_query_params(self, params: dict) -> dict: """Build query parameters from tool parameters.""" query_params = {} # Simple string parameters for key in ["query", "tag", "type", "limit", "page"]: if params.get(key): query_params[key] = params[key] # Comma-separated list parameters for key in ["dashboardIds", "dashboardUIDs", "folderUIDs"]: if params.get(key): values = params[key].split(",") query_params[key] = [v.strip() for v in values if v.strip()] # Boolean parameters if params.get("starred") is not None: query_params["starred"] = str(params["starred"]).lower() return query_params def _invoke(self, params: dict, context: ToolInvokeContext) -> StructuredToolResult: query_params = self._build_query_params(params) return self._make_grafana_request("/api/search", params, query_params)
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (3)
holmes/core/tools.py(1 hunks)holmes/plugins/toolsets/grafana/toolset_grafana.py(2 hunks)holmes/plugins/toolsets/grafana/toolset_grafana_dashboard.jinja2(1 hunks)
🧰 Additional context used
📓 Path-based instructions (2)
holmes/plugins/toolsets/**/*
📄 CodeRabbit inference engine (CLAUDE.md)
Toolsets must be located as holmes/plugins/toolsets/{name}.yaml or holmes/plugins/toolsets/{name}/
Files:
holmes/plugins/toolsets/grafana/toolset_grafana_dashboard.jinja2holmes/plugins/toolsets/grafana/toolset_grafana.py
**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
**/*.py: Use Ruff for formatting and linting (configured in pyproject.toml) for all Python code
Type hints are required; code should pass mypy (configured in pyproject.toml)
ALWAYS place Python imports at the top of the file, not inside functions or methods
Files:
holmes/core/tools.pyholmes/plugins/toolsets/grafana/toolset_grafana.py
🧬 Code graph analysis (1)
holmes/plugins/toolsets/grafana/toolset_grafana.py (4)
holmes/plugins/toolsets/grafana/common.py (2)
get_base_url(52-56)GrafanaConfig(9-22)holmes/plugins/toolsets/utils.py (1)
toolset_name_for_one_liner(232-236)holmes/plugins/toolsets/grafana/base_grafana_toolset.py (1)
BaseGrafanaToolset(11-54)holmes/core/tools.py (9)
_load_llm_instructions_from_file(770-778)Tool(173-365)StructuredToolResult(79-103)ToolParameter(155-161)_invoke(352-361)_invoke(414-446)ToolInvokeContext(164-170)get_parameterized_one_liner(364-365)get_parameterized_one_liner(391-398)
🪛 Ruff (0.14.1)
holmes/plugins/toolsets/grafana/toolset_grafana.py
74-74: Probable use of requests call without timeout
(S113)
141-141: Unused method argument: context
(ARG002)
178-178: Unused method argument: params
(ARG002)
197-197: Unused method argument: context
(ARG002)
214-214: Unused method argument: context
(ARG002)
217-217: Unused method argument: params
(ARG002)
230-230: Unused method argument: context
(ARG002)
233-233: Unused method argument: params
(ARG002)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (3)
- GitHub Check: Pre-commit checks
- GitHub Check: llm_evals
- GitHub Check: build
🔇 Additional comments (5)
holmes/core/tools.py (1)
770-778: LGTM! Clean helper method for file-based LLM instruction loading.The implementation correctly constructs an absolute path and delegates to the existing
_load_llm_instructionsmethod using afile://URL scheme. The method has clear documentation and proper type hints.holmes/plugins/toolsets/grafana/toolset_grafana_dashboard.jinja2 (1)
1-26: LGTM! Clear and structured Prometheus query execution guidance.The template provides well-organized instructions for running Prometheus queries from Grafana dashboards, including key rules, variable substitution guidelines, and a concrete example. The documentation will help ensure consistent query execution behavior.
holmes/plugins/toolsets/grafana/toolset_grafana.py (3)
19-23: LGTM! Appropriate config specialization for dashboard toolset.The override of
healthcheckto"api/health"is intentional and well-documented, differentiating the dashboard-specific configuration from the base Grafana configuration.
25-49: LGTM! Well-structured toolset initialization.The toolset properly initializes the base class, loads LLM instructions from the template file using the new helper method, and provides typed access to the configuration through the
grafana_configproperty.
182-234: LGTM! Clean and straightforward tool implementations.The three dashboard tools (
GetDashboardByUID,GetHomeDashboard,GetDashboardTags) are well-implemented with proper initialization, simple delegation to the shared request helper, and appropriate user-facing descriptions.Note: The static analysis warnings about unused
contextandparamsparameters are false positives—these parameters are required by the abstract method signatures in the baseToolclass.
Sheeproid
force-pushed
the
grafana-dashboard-ts
branch
from
a9066ae to
738a81e
Compare
There was a problem hiding this comment.
Actionable comments posted: 1
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (2)
docs/data-sources/builtin-toolsets/grafanadashboards.md(1 hunks)holmes/plugins/toolsets/grafana/toolset_grafana.py(2 hunks)
🧰 Additional context used
📓 Path-based instructions (3)
**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
**/*.py: Use Ruff for formatting and linting (configured in pyproject.toml) for all Python code
Type hints are required; code should pass mypy (configured in pyproject.toml)
ALWAYS place Python imports at the top of the file, not inside functions or methods
Files:
holmes/plugins/toolsets/grafana/toolset_grafana.py
holmes/plugins/toolsets/**/*
📄 CodeRabbit inference engine (CLAUDE.md)
Toolsets must be located as holmes/plugins/toolsets/{name}.yaml or holmes/plugins/toolsets/{name}/
Files:
holmes/plugins/toolsets/grafana/toolset_grafana.py
docs/**/*.md
📄 CodeRabbit inference engine (CLAUDE.md)
In MkDocs content, always add a blank line between a header or bold text and a following list so lists render correctly
Files:
docs/data-sources/builtin-toolsets/grafanadashboards.md
🧬 Code graph analysis (1)
holmes/plugins/toolsets/grafana/toolset_grafana.py (4)
holmes/plugins/toolsets/grafana/common.py (3)
get_base_url(52-56)GrafanaConfig(9-22)build_headers(25-36)holmes/plugins/toolsets/utils.py (1)
toolset_name_for_one_liner(232-236)holmes/plugins/toolsets/grafana/base_grafana_toolset.py (1)
BaseGrafanaToolset(11-54)holmes/core/tools.py (10)
_load_llm_instructions_from_file(770-778)Tool(173-365)StructuredToolResult(79-103)StructuredToolResultStatus(52-76)ToolParameter(155-161)_invoke(352-361)_invoke(414-446)ToolInvokeContext(164-170)get_parameterized_one_liner(364-365)get_parameterized_one_liner(391-398)
🪛 LanguageTool
docs/data-sources/builtin-toolsets/grafanadashboards.md
[style] ~3-~3: Consider a different adjective to strengthen your wording.
Context: ...nd extraction of Prometheus queries for deeper analysis. ## Prerequisites A [Grafana...
(DEEP_PROFOUND)
🪛 markdownlint-cli2 (0.18.1)
docs/data-sources/builtin-toolsets/grafanadashboards.md
76-76: Code block style
Expected: indented; Actual: fenced
(MD046, code-block-style)
81-81: Code block style
Expected: indented; Actual: fenced
(MD046, code-block-style)
86-86: Code block style
Expected: indented; Actual: fenced
(MD046, code-block-style)
🪛 Ruff (0.14.1)
holmes/plugins/toolsets/grafana/toolset_grafana.py
81-81: Probable use of requests call without timeout
(S113)
148-148: Unused method argument: context
(ARG002)
185-185: Unused method argument: params
(ARG002)
204-204: Unused method argument: context
(ARG002)
221-221: Unused method argument: context
(ARG002)
224-224: Unused method argument: params
(ARG002)
237-237: Unused method argument: context
(ARG002)
240-240: Unused method argument: params
(ARG002)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (3)
- GitHub Check: llm_evals
- GitHub Check: Pre-commit checks
- GitHub Check: build
🔇 Additional comments (7)
holmes/plugins/toolsets/grafana/toolset_grafana.py (6)
23-26: LGTM!The config class correctly extends
GrafanaConfigand appropriately overrides the healthcheck endpoint toapi/healthfor dashboard-specific health checks.
29-52: LGTM!The toolset is well-structured:
- Config class properly wired to
GrafanaDashboardConfig- All four tools correctly instantiated
- Proper use of the
_load_llm_instructions_from_filehelper for loading LLM instructions from the Jinja2 template- Type-safe access to config via the property
93-186: LGTM!The SearchDashboards tool is well-implemented:
- Comprehensive parameter definitions matching the Grafana API
- Correct handling of comma-separated strings for list parameters (dashboardIds, dashboardUIDs, folderUIDs)
- Proper boolean-to-string conversion for the API
- Clean delegation to
_make_grafana_requestNote: Ruff's warnings about unused
contextandparamsare false positives—these parameters are required by the abstract method signatures from the baseToolclass.
189-209: LGTM!Clean and focused implementation for retrieving a dashboard by UID. The endpoint construction is correct and the one-liner includes the UID for better traceability.
212-225: LGTM!Straightforward implementation for retrieving the home dashboard. No parameters needed, clean delegation to the base request method.
228-241: LGTM!Clean implementation for retrieving dashboard tags. Follows the same pattern as GetHomeDashboard with no parameters required.
docs/data-sources/builtin-toolsets/grafanadashboards.md (1)
1-88: LGTM!Excellent documentation for the Grafana Dashboards integration:
- Clear prerequisites and configuration instructions for both CLI and Helm
- Accurate capabilities table matching the four implemented tools
- Helpful examples demonstrating practical use cases
- Proper formatting with blank lines before lists as per coding guidelines
Note: The static analysis hints about code block style (fenced vs. indented) and the word "deeper" are false positives—fenced code blocks are the modern standard and more readable.
No description provided.