Route queries to the best opencli search source based on the topic and scenario. The core goal of this skill is not to memorize commands, but to first locate the data source, and then have the Agent read the real-time help itself via opencli , to avoid documentation drift.
Mandatory Pre-check
Before each use, you must first do the following two steps:
- Run
opencli list -f yaml - Use the live registry to confirm whether the candidate site exists, and check
strategy、browser、domain
After selecting a site, you must then do the following two steps:
- Run
opencli <site> -hto see which subcommands the site has - If a subcommand has been pinned down, then run
opencli <site> <command> -hto see the parameters, output columns, and policies
Do not hardcode parameters or assume command signatures in the skill documentation; rely on the real-time output of opencli ... -h .
Main Routing Rules
Use only this one rule; do not maintain multiple sets of priorities:
- When the user explicitly specifies a website, platform, or data source, use the corresponding website directly.
- When the user does not specify a website, prefer selecting only one AI source:
grok、doubao、geminichoose one of the three. - When the AI returns insufficient content, lacks raw data, needs authoritative corroboration, or needs vertical results, then supplement with 1-2 specialized sources.
Per-question budget and rate limits
Understand a "single user question" as one instance of problem-solving under the same intent chain; for follow-up questions, clarifications, or additional conditions in the same round, if the core question has not changed, it still counts as the same question.
First establish a site call ledger. After each actual execution of a search command, update it immediately:
sitequerycountstatus
Counting rules:
opencli list -f yaml、opencli <site> -h、opencli <site> <command> -hbelongs to pre-check and help, and is not counted as a search count- One actual
opencli <site> ...search/query execution counts as 1 call for that site - If the same site fails due to an error, timeout, CAPTCHA, anti-scraping, or abnormal login state, it also counts as 1 call; do not retry indefinitely
Rate limits:
- Hard limit for AI sites: within the same question, each AI site may be called at most 1 time
- The default strategy is still to select only 1 AI site; do not chain multiple AI sites into a regular process
- Only when the user explicitly requests a comparison of multiple AI sites may other AI sites be called additionally; but each named AI site is still limited to at most 1 call
- Non-AI sites are limited to at most 2 calls by default
- The second call to a non-AI site must have a clear reason, for example the first result was too broad and needs to be narrowed by time, region, category, sorting, or keywords
- Do not make a third call to a non-AI site; if the information is still insufficient, stop expanding the search and clearly state the gap
Handling after rate limiting is triggered:
- Record: "Skipped: reached rate limit"
- Prefer switching to another site of the same type
- If there is no suitable alternative source, answer directly based on the information already collected, and explain the coverage and gaps
Query-ending report
After each query ends, a brief "search summary" must be appended to the end of the answer, containing at least the following three items:
- What websites were used for searching
- What keywords were searched on each website
- How many times each website was searched
If any site was skipped due to rate limiting, that must also be clearly stated.
It is recommended to use the following fixed format:
搜索摘要
- 网站:<site1> | 查询词:<term1> | 次数:<n>
- 网站:<site2> | 查询词:<term2>;<term3> | 次数:<n>
- 已跳过:<site3>,原因:达到频率上限
AI source selection
grokSuitable for real-time discussions, English-language internet public opinion, Twitter/X contexts, and trending topic tracking.doubaoSuitable for Chinese-language contexts, the ByteDance Douyin ecosystem, lifestyle content, Chinese trending topics, and general Chinese-language Q&A.geminiSuitable for global web pages, English-language materials, general information retrieval, and background overviews.
If the user has not specified a website, first determine the language and context by default, then choose only one from these three.
Once a given AI site has already performed one real query, do not rewrite the keywords and call that same AI site again for the same question. If the answer is insufficient, prioritize supplementing with dedicated sources rather than repeatedly hammering the same AI site.
AI query keyword suggestions
When using an AI source, do not just throw in a single overly short keyword. Prioritize constructing a query as "topic + goal + constraints".
- Topic The object, event, product, person, company, or technical term the user actually wants to look up.
- Goal What kind of result is wanted, such as a summary, comparison, reason, trend, recommendation, or original lead.
- Constraints Language, region, time range, platform scope, audience, price range, job location, and whether original sources need to be cited.
Prioritize using the following style of expression:
<主题> + <你要回答的问题><主题> + <时间范围/地区/语言><主题> + <平台或来源范围><主题> + <输出要求>
Avoid entering only:
- A single noun
- Trending questions without a time range
- Shopping, job-seeking, or travel questions without a region restriction
- Social media questions without a platform restriction
When to supplement with dedicated sources
When any of the following situations occurs, then supplement with dedicated sources:
- The AI gives a summary, but you need the original post, original video, original product, or original job listing result
- The AI's coverage is insufficient and omits information from vertical sites
- Higher authority or stronger domain relevance is needed
- The user explicitly requests "find it from a certain platform"
A single query is usually kept to 1 AI source + 1 to 2 dedicated sources, to avoid result overload.
Handling unavailable sources
When a site is unavailable:
- Do not abort the entire search just because a single source fails
- Record: "Skipped: unavailable"
- Fall back to other sites of the same type, or fall back to an AI source
- Always base on the actual results of
opencli list -f yamlandopencli <site> -hThe actual result shall prevail
Do not assume any site is "absolutely available". Even for public sites, base on the live help and execution results in the current environment.
Reference files
Read the corresponding files as needed:
references/sources-ai.md— AI default sourcesreferences/sources-tech.md— Technology / Academicreferences/sources-social.md— Social mediareferences/sources-media.md— Media / Entertainmentreferences/sources-info.md— News / Knowledgereferences/sources-shopping.md— Shoppingreferences/sources-travel.md— Travelreferences/sources-other.md— Other vertical sources
Only read the files relevant to the current query; there is no need to load all of them.