Add page-precise URL exclusion to Tavily search tools#1496
Open
jubick1337 wants to merge 3 commits into
Open
Conversation
Lets a blocklist target a single page/paper (e.g. one arxiv id) without banning the whole domain. url_substring patterns load from the same exclude config as domains and apply to both search results and page access. Signed-off-by: mnovikov <mnovikov@nvidia.com>
Contributor
📝 WalkthroughWalkthroughTavily exclude-config handling now supports ChangesURL substring exclusion
Sequence Diagram(s)sequenceDiagram
participant ConfigFile
participant DirectTavilyBase
participant DirectTavilyGymTool
participant DirectTavilyBrowserTool
ConfigFile->>DirectTavilyBase: load exclude config JSON
DirectTavilyBase->>DirectTavilyBase: parse domain and url_substring blocks
DirectTavilyBase->>DirectTavilyBase: store _exclude_domains and _exclude_url_substrings
DirectTavilyGymTool->>DirectTavilyBase: _is_url_blocked(url)
DirectTavilyBase-->>DirectTavilyGymTool: blocked or allowed
DirectTavilyGymTool->>DirectTavilyGymTool: filter search results / reject page lookup / reject scroll
DirectTavilyBrowserTool->>DirectTavilyBase: _is_url_blocked(url)
DirectTavilyBase-->>DirectTavilyBrowserTool: blocked or allowed
DirectTavilyBrowserTool->>DirectTavilyBrowserTool: skip search-memory entry / stop page extraction
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes 🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
🧹 Nitpick comments (1)
nemo_skills/mcp/servers/tavily_search_tool.py (1)
202-219: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick winConsider a single parse/load pass for domains and substrings.
_parse_exclude_url_substrings/_load_exclude_url_substrings_from_fileduplicate the domain variants, andconfigurereads the sameexclude_domains_configfile twice (once for each type). A single loader that walksnotices/propertiesonce and returns both lists keeps the code DRY and avoids the redundant parse.As per coding guidelines: "Keep code simple and elegant; reuse/extend existing functionality when possible."♻️ Sketch
def _parse_exclude_config(exclude_config: dict[str, Any]) -> tuple[list[str], list[str]]: domains, substrings = [], [] for notice in exclude_config["notices"]: for prop in notice["properties"]: if prop["type"] == "domain": domains.append(prop["value"]) elif prop["type"] == "url_substring": substrings.append(prop["value"]) return domains, substringsThen
configureopens the file once and unpacks both lists.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@nemo_skills/mcp/servers/tavily_search_tool.py` around lines 202 - 219, The exclude config loading is duplicated between the domain and URL-substring paths, and configure currently parses the same file twice. Refactor the existing _parse_exclude_url_substrings and the corresponding domain parser into a single shared parser/loader that walks exclude_config["notices"] and their properties once, collecting both "domain" and "url_substring" values. Then update configure and the existing _load_exclude_*_from_file helpers to reuse that shared logic and return both lists from one file read.Source: Coding guidelines
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Nitpick comments:
In `@nemo_skills/mcp/servers/tavily_search_tool.py`:
- Around line 202-219: The exclude config loading is duplicated between the
domain and URL-substring paths, and configure currently parses the same file
twice. Refactor the existing _parse_exclude_url_substrings and the corresponding
domain parser into a single shared parser/loader that walks
exclude_config["notices"] and their properties once, collecting both "domain"
and "url_substring" values. Then update configure and the existing
_load_exclude_*_from_file helpers to reuse that shared logic and return both
lists from one file read.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Enterprise
Run ID: 76c30bf1-0fec-4d2c-aeeb-c04bde1ce226
📒 Files selected for processing (2)
nemo_skills/mcp/servers/tavily_search_tool.pytests/test_tavily_search_tool.py
Ships exclude_domains_hle_opus.json (9 whole-domain + 31 page-precise url_substring bans) next to the Tavily tools as a ready-made exclude_domains_config, cited to the Claude Opus 4.8 System Card and referenced from the docstring. Whitelisted in .gitignore since the repo blanket-ignores *.json. Signed-off-by: mnovikov <mnovikov@nvidia.com>
Contributor
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@nemo_skills/mcp/servers/exclude_domains_hle_opus.json`:
- Line 42: The Medium blocklist entry in exclude_domains_hle_opus.json has a
slug typo, so the intended URL will not be matched by the url_substring filter.
Update the value in the relevant {"type": "url_substring"} entry to use the
correct Medium slug, and verify the exclusion still targets the intended page
based on the existing substring-matching behavior.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Enterprise
Run ID: 3c9be89a-6b96-4ca6-9dfd-d425e94e1fe3
📒 Files selected for processing (3)
.gitignorenemo_skills/mcp/servers/exclude_domains_hle_opus.jsonnemo_skills/mcp/servers/tavily_search_tool.py
✅ Files skipped from review due to trivial changes (1)
- .gitignore
🚧 Files skipped from review as they are similar to previous changes (1)
- nemo_skills/mcp/servers/tavily_search_tool.py
| {"type": "url_substring", "value": "github.com/hanjanghoon/DEER"}, | ||
| {"type": "url_substring", "value": "github.com/repos/hanjanghoon/DEER"}, | ||
| {"type": "url_substring", "value": "mveteanu/HLE_PDF"}, | ||
| {"type": "url_substring", "value": "medium.com/@82deutschmark/o3-quiet-breakthrough-1bf9f0fobafc84"}, |
Contributor
There was a problem hiding this comment.
🎯 Functional Correctness | 🟠 Major | ⚡ Quick win
Fix the Medium slug typo so the intended page is actually excluded.
Line 42 has 1bf9f0fobafc84, but indexed source excerpts list this blocklist URL as medium.com/@82deutschmark/o3-quiet-breakthrough-1bf9f0bafc84; because substring matching only lowercases and removes /, the current typo will not match the intended URL. (studylib.net)
Proposed fix
- {"type": "url_substring", "value": "medium.com/@82deutschmark/o3-quiet-breakthrough-1bf9f0fobafc84"},
+ {"type": "url_substring", "value": "medium.com/@82deutschmark/o3-quiet-breakthrough-1bf9f0bafc84"},📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| {"type": "url_substring", "value": "medium.com/@82deutschmark/o3-quiet-breakthrough-1bf9f0fobafc84"}, | |
| {"type": "url_substring", "value": "medium.com/@82deutschmark/o3-quiet-breakthrough-1bf9f0bafc84"}, |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@nemo_skills/mcp/servers/exclude_domains_hle_opus.json` at line 42, The Medium
blocklist entry in exclude_domains_hle_opus.json has a slug typo, so the
intended URL will not be matched by the url_substring filter. Update the value
in the relevant {"type": "url_substring"} entry to use the correct Medium slug,
and verify the exclusion still targets the intended page based on the existing
substring-matching behavior.
Collaborator
|
lgtm |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Lets a blocklist target a single page/paper (e.g. one arxiv id) without banning the whole domain. url_substring patterns load from the same exclude config as domains and apply to both search results and page access.
Summary by CodeRabbit