Google Scholar

Empower your agents to search academic publications, citations, and research papers through Google Scholar, powered by SerpApi.

This guide will walk you through obtaining a SerpApi key, configuring the SVAHNAR tool, and building research discovery workflows.

💡 Core Concepts

To configure this tool effectively, you need to understand the underlying capabilities, the two distinct search modes, and the available filtering parameters.

1. What can this tool do?

The Google Scholar tool queries Google Scholar via SerpApi to search academic publications, journals, conference papers, theses, and books indexed across the web.

Capability	Description
Keyword search	Search by topic, keywords, author name, or journal title.
Citation lookup	Find all papers that cite a specific article using its unique ID.
Year range filtering	Restrict results to papers published within a specific year window.
Sort by recency	Surface the most recently published papers first.
Patent exclusion	Filter out patents and legal documents from results.
Pagination	Page through large result sets using the `start` offset.
Language localization	Return results in a specific language.

note

This tool is read-only. It searches and returns publication metadata — it does not download full-text papers, access paywalled content, or interact with journal databases directly.

2. Authentication

This tool uses a SerpApi API Key — a static key passed with every request.

No OAuth required: Generate a key once from the SerpApi dashboard and paste it into SVAHNAR. No per-user login flow needed.
Rate limits: SerpApi enforces monthly search credit limits based on your plan. Each tool call consumes one search credit.
Maintenance: API keys do not expire automatically. They are invalidated only if manually rotated from the SerpApi dashboard.

tip

If you already use Google Jobs, Google Finance, Google Flights, or Google Hotels tools, your SerpApi key is the same — one key works across all SerpApi-powered tools in SVAHNAR.

3. Two Search Modes

The tool operates in two distinct modes depending on whether you use q or cites:

Mode	Key Parameter	Use Case
Keyword Search	`q`	Find papers by topic, author, journal, or any keyword.
Cited By Search	`cites`	Find all papers that have cited a specific article by its Scholar ID.

caution

q and cites serve different purposes and should not be used simultaneously. Use q for discovery — use cites for tracing how a specific paper has influenced the field.

4. The `scisbd` Sort Parameter

The scisbd field controls how results are sorted and what content is included:

Value	Behavior
`0`	Sort by relevance (default). Best for general keyword searches.
`1`	Sort by date — abstracts only. Most recently published papers first.
`2`	Sort by date — everything. Most recent, including non-abstract results.

tip

Use scisbd: 1 combined with as_ylo and as_yhi when you want the most recent peer-reviewed papers on a topic. Use scisbd: 0 when you want the most authoritative and widely-cited results regardless of date.

5. Parameter Reference

Parameter	Type	Required	Description	Example
`q`	string	Yes (unless using `cites`)	Search query — keywords, author name, or journal.	`"machine learning"`, `"author:Hinton"`
`engine`	string	No	Must be `"google_scholar"` when specified.	`"google_scholar"`
`cites`	string	No	Scholar article ID — triggers a "Cited By" search for that paper.	`"127598..."`
`as_ylo`	int	No	Start year for date range filter (inclusive).	`2020`
`as_yhi`	int	No	End year for date range filter (inclusive).	`2024`
`scisbd`	int	No	Sort mode. `0` = Relevance, `1` = Date (abstracts), `2` = Date (all).	`0`
`as_sdt`	string	No	Search type filter. `"0"` excludes patents. `"1"` includes only case law.	`"0"`
`num`	int	No	Number of results to return. Range: 1–20. Default: 10.	`10`
`start`	int	No	Pagination offset. Use `0`, `10`, `20`... to page through results.	`10`
`hl`	string	No	2-letter language code for results.	`"en"`

Common as_sdt Values:

Value	Filter Applied
`"0"`	Exclude patents — return academic papers only.
`"1"`	Case law only.
`"5"`	Include patents in results.
`"0,5"`	All results — articles and patents.

🔑 Prerequisites

Create a SerpApi Account

Go to https://serpapi.com and sign up for an account.
SerpApi offers a free tier with 100 searches/month — sufficient for development and light research agent usage.
For research-heavy workloads or automated literature monitoring, select a paid plan based on your expected monthly query volume.

Get Your API Key

After signing in, go to your SerpApi Dashboard.
Copy the API Key shown on the dashboard.

caution

Never commit this key to version control or hardcode it in config files. Use SVAHNAR Key Vault (${serpapi_key}) to reference it safely. If you are already using other SerpApi tools in SVAHNAR, the same key applies — no new account needed.

⚙️ Configuration Steps

Add the Tool in SVAHNAR

Open your SVAHNAR Agent Configuration.
Add the Google Scholar tool and enter your SerpApi credentials:
- api_key — your SerpApi API key
Save the configuration.

Verify the Connection

To confirm your API key is working:

Trigger a test agent run with a simple payload:

{
  "q": "machine learning",
  "hl": "en",
  "num": 5
}

A valid response will return a list of academic papers with titles, authors, publication year, and citation counts.
If you receive an Invalid API key error, verify the key was copied in full from your SerpApi dashboard.
If you receive a credits exhausted error, your monthly search quota has been reached — upgrade your SerpApi plan or wait for the next billing cycle.

📚 Practical Recipes (Examples)

Recipe 1: Literature Review Agent

Use Case: An agent that finds the most relevant and highly-cited papers on a research topic within a specific time window.

create_vertical_agent_network:
  agent-1:
    agent_name: literature_review_agent
    LLM_config:
        params:
          model: gpt-4o
    tools:
      tool_assigned:
        - name: GoogleScholar
          config:
            api_key: ${serpapi_key}
    agent_function:
      - You are a research literature assistant.
      - Ask the user for their research topic and preferred year range (e.g., 2019–2024).
      - Set 'q' to the topic keywords, 'as_ylo' and 'as_yhi' to the year range, and 'scisbd' to 0 (relevance) to surface the most authoritative papers first.
      - Set 'as_sdt' to '0' to exclude patents and return academic papers only.
      - Set 'num' to 10 for a comprehensive first-page result set.
      - Return a structured list of results — paper title, authors, publication year, journal, and citation count.
    incoming_edge:
      - Start
    outgoing_edge: []

Recipe 2: Recent Publications Monitor Agent

Use Case: An agent that surfaces the latest published papers on a topic, sorted by recency.

create_vertical_agent_network:
  agent-1:
    agent_name: recent_publications_agent
    LLM_config:
        params:
          model: gpt-4o
    tools:
      tool_assigned:
        - name: GoogleScholar
          config:
            api_key: ${serpapi_key}
    agent_function:
      - You are an academic research monitor.
      - Use the configured research topic as 'q', set 'scisbd' to 1 (sort by date, abstracts only) to surface the most recently published papers.
      - Set 'as_ylo' to the current year to restrict results to the latest publications only.
      - Set 'as_sdt' to '0' to exclude patents.
      - Return paper titles, authors, year, and a brief summary of the abstract where available.
      - If the user wants to go deeper, use 'start' to paginate — set to 10 for page 2, 20 for page 3.
    incoming_edge:
      - Start
    outgoing_edge: []

Recipe 3: Citation Tracer Agent

Use Case: An agent that finds all papers citing a specific foundational paper — useful for tracking how a seminal work has influenced the field.

create_vertical_agent_network:
  agent-1:
    agent_name: citation_tracer_agent
    LLM_config:
        params:
          model: gpt-4o
    tools:
      tool_assigned:
        - name: GoogleScholar
          config:
            api_key: ${serpapi_key}
    agent_function:
      - You are a citation analysis assistant.
      - Ask the user for the Google Scholar article ID of the paper they want to trace (found in the Scholar URL or a prior keyword search result).
      - Use 'cites' with that article ID — do NOT set 'q' simultaneously.
      - Set 'scisbd' to 0 (relevance) to surface the most impactful citing papers first, or 1 (date) to find the most recent work building on it.
      - Set 'num' to 20 for a wide first pass.
      - Return the citing papers grouped by year — most recent first — with title, authors, and their own citation count.
    incoming_edge:
      - Start
    outgoing_edge: []

💡 Tip: SVAHNAR Key Vault

Never hardcode your api_key in plain text files. Use SVAHNAR Key Vault references (e.g., ${serpapi_key}) to keep credentials secure. The same SerpApi key used for Google Jobs, Finance, Flights, and Hotels works here — no separate credential needed.

💡 Tip: Finding a Paper's Scholar ID for `cites`

To get the cites ID for a paper:

Search for the paper using a keyword q query first.
In the response, each result includes a result_id or a cites link — this alphanumeric string is the Scholar ID.
Pass that string as the cites value in a subsequent query to fetch all papers citing it.

🚑 Troubleshooting

Invalid API key Error
- Your SerpApi key is incorrect or was not copied in full. Go to your SerpApi Dashboard, copy the key again, and update it in SVAHNAR Key Vault.
Credits Exhausted Error
- Your monthly SerpApi search quota has been reached. Each Google Scholar query consumes one search credit.
- Upgrade at serpapi.com/pricing. Reduce consumption by increasing num to get more results per query rather than paginating with many small calls.
Results Include Patents When Not Wanted
- Set as_sdt: "0" to exclude patents from results. Without this, Google Scholar may return patent filings alongside academic papers.
cites Search Returning Unrelated Results
- The cites ID must be the exact Scholar article ID — a long numeric string. If it came from a URL, ensure no extra characters (slashes, query strings) were included.
- Do not combine cites with q — Scholar treats these as separate modes. Using both simultaneously produces unpredictable results.
Year Range Filter Not Working
- Ensure as_ylo and as_yhi are passed as integers — not strings. as_ylo: 2020 is correct; as_ylo: "2020" may be ignored.
- Both fields are inclusive — as_ylo: 2020, as_yhi: 2024 returns papers from 2020 through 2024.
num Exceeding 20 Returns an Error
- Google Scholar limits results to a maximum of 20 per query. To retrieve more results, use start to paginate — start: 0 for page 1, start: 10 for page 2, start: 20 for page 3, and so on.
Results in Wrong Language
- Set hl to the desired 2-letter language code — e.g., "en" for English, "hi" for Hindi. Without it, Scholar may return results in the language of the most relevant indexed papers, which may not match the user's expectation.
engine Field Causing Errors
- If specified, engine must be exactly "google_scholar". Any other value routes to a different SerpApi engine. When in doubt, omit the field entirely — it defaults correctly.

💡 Core Concepts​

1. What can this tool do?​

2. Authentication​

3. Two Search Modes​

4. The scisbd Sort Parameter​

5. Parameter Reference​

🔑 Prerequisites​

Create a SerpApi Account​

Get Your API Key​

⚙️ Configuration Steps​

Add the Tool in SVAHNAR​

Verify the Connection​

📚 Practical Recipes (Examples)​

Recipe 1: Literature Review Agent​

Recipe 2: Recent Publications Monitor Agent​

Recipe 3: Citation Tracer Agent​

💡 Tip: SVAHNAR Key Vault​

💡 Tip: Finding a Paper's Scholar ID for cites​

🚑 Troubleshooting​