"""Pre-built prompt workflows for common ADS tasks."""
from __future__ import annotations
from typing import Annotated
from pydantic import Field
from mcp_server_ads.server import mcp
@mcp.prompt()
def literature_review(
topic: Annotated[str, Field(description="Research topic to review")],
year_range: Annotated[
str,
Field(description="Year range (e.g. '2018-2025'). Default: '2020-2025'"),
] = "2020-2025",
) -> str:
"""Multi-step literature review workflow for a research topic."""
return f"""\
Conduct a comprehensive literature review on "{topic}" using the ADS database.
Follow these steps:
1. **Initial Search**: Use ads_search to find recent papers on this topic:
- Query: abs:"{topic}" year:{year_range} property:refereed
- Sort by citation_count desc, get 20 results
2. **Identify Key Papers**: From the results, identify the 5-10 most influential papers
(highest citation count, most relevant to the topic).
3. **Expand Search**: For the top 2-3 papers, use ads_search with the citations()
functional operator to find papers that cite them, revealing recent follow-up work.
4. **Compute Metrics**: Use ads_metrics on the key papers to understand the field's
impact and growth.
5. **Export Bibliography**: Use ads_export to export the final curated list in BibTeX format.
6. **Summarize**: Provide a structured summary including:
- Key findings and trends
- Most influential papers and authors
- Recent developments
- Gaps or open questions
Topic: {topic}
Year range: {year_range}
"""
@mcp.prompt()
def citation_analysis(
bibcodes: Annotated[
str,
Field(description="Comma-separated list of bibcodes to analyze"),
],
) -> str:
"""Citation network analysis workflow for a set of papers."""
bibcode_list = [b.strip() for b in bibcodes.split(",")]
formatted = "\n".join(f"- `{b}`" for b in bibcode_list)
return f"""\
Perform a citation network analysis on the following papers:
{formatted}
Follow these steps:
1. **Paper Details**: Use ads_bigquery to retrieve full metadata for all papers.
2. **Citation Metrics**: Use ads_metrics to compute citation statistics including
h-index, citation counts, and read counts.
3. **Citation Network**: Use ads_network with type="paper" to visualize the paper network
and identify clusters.
4. **Author Network**: Use ads_network with type="author" to identify key collaborator groups.
5. **Citation Suggestions**: Use ads_citation_helper to find related papers
that should be cited.
6. **Summarize**: Provide analysis including:
- Citation count distribution
- Key metric indicators (h-index, etc.)
- Network clusters and their themes
- Suggested additional references
"""
@mcp.prompt()
def generate_bibliography(
source: Annotated[
str,
Field(
description="Either a search query or comma-separated bibcodes"
),
],
format: Annotated[
str,
Field(description="Export format (bibtex, aastex, ris, etc.). Default: bibtex"),
] = "bibtex",
) -> str:
"""Generate a formatted bibliography from a search or bibcode list."""
return f"""\
Generate a formatted bibliography from the following source.
Source: {source}
Steps:
1. **Resolve Papers**:
- If the source looks like bibcodes (contains slashes or dots typical of bibcodes),
use ads_bigquery to verify them.
- Otherwise, use ads_search to find matching papers.
2. **Review Results**: Show the papers found and confirm they are the intended ones.
3. **Export**: Use ads_export with format="{format}" to generate the bibliography.
4. **Deliver**: Return the formatted bibliography ready for use.
Requested format: {format}
"""
