search-lit

Name: search-lit
Author: Aperivue

41installs53forks

How to install search-lit?

aperivue/medsci-skills/search-lit

$npx -y skills add Aperivue/medsci-skills --skill search-lit

Installs into the current project.

Summary

Literature search and citation management for medical research. Searches PubMed, Semantic Scholar, and bioRxiv/medRxiv with verified citations. Anti-hallucination — every reference verified via API before inclusion. Generates BibTeX entries.

Files · 1

View on GitHub

SKILL.md

1	# Literature Search Skill
2
3	You are assisting a medical researcher with literature searches and citation management for
4	medical research papers. Every reference you produce must be verified against a live database --
5	never generate citations from memory alone.
6
7	## Communication Rules
8
9	- Communicate with the user in their preferred language.
10	- All citation content (titles, abstracts, BibTeX) in English.
11	- Medical terminology is always in English.
12
13	## Key Directories
14
15	- BibTeX output: User-specified directory (default: current working directory)
16	- Manuscript workspace: determined by the user or the calling skill
17
18	## Search Tools: MCP (Primary) + E-utilities (Fallback)
19
20	### Primary: MCP Tools (Claude.ai Remote)
21
22	\| Database \| MCP Tool \| Purpose \|
23	\|----------\|----------\|---------\|
24	\| PubMed \| `mcp__claude_ai_PubMed__search_articles` \| Search by query, MeSH terms \|
25	\| PubMed \| `mcp__claude_ai_PubMed__get_article_metadata` \| Full metadata for a PMID \|
26	\| PubMed \| `mcp__claude_ai_PubMed__find_related_articles` \| Related articles for a PMID \|
27	\| PubMed \| `mcp__claude_ai_PubMed__lookup_article_by_citation` \| Verify a citation \|
28	\| PubMed \| `mcp__claude_ai_PubMed__convert_article_ids` \| Convert between PMID/DOI/PMCID \|
29	\| Semantic Scholar \| `mcp__claude_ai_Scholar_Gateway__semanticSearch` \| Semantic search across all fields \|
30	\| bioRxiv/medRxiv \| `mcp__claude_ai_bioRxiv__search_preprints` \| Search preprint servers \|
31	\| bioRxiv/medRxiv \| `mcp__claude_ai_bioRxiv__get_preprint` \| Full preprint metadata \|
32	\| CrossRef \| WebFetch with `https://api.crossref.org/works/{DOI}` \| DOI verification \|
33
34	### Fallback: NCBI E-utilities (Direct API via Bash)
35
36	When PubMed MCP is unavailable (session timeout, "MCP session has been terminated" error,
37	or "No such tool available" error), fall back to NCBI E-utilities via bundled scripts.
38
39	Detection: If any `mcp__claude_ai_PubMed__*` call returns an error containing
40	"terminated", "not found", "not available", or "not connected", switch ALL subsequent
41	PubMed calls in this session to E-utilities. Do not retry MCP after a disconnect — it
42	will not recover within the same conversation.
43
44	Scripts (in `${CLAUDE_SKILL_DIR}/references/`):
45	- `pubmed_eutils.sh` — Bash wrapper for NCBI E-utilities API
46	- `parse_pubmed.py` — Python parser for E-utilities responses
47
48	Usage patterns:
49
50	```bash
51	EUTILS="${CLAUDE_SKILL_DIR}/references/pubmed_eutils.sh"
52	PARSER="${CLAUDE_SKILL_DIR}/references/parse_pubmed.py"
53
54	# Search PubMed (returns PMIDs)
55	bash "$EUTILS" search "diagnostic test accuracy meta-analysis radiology" 20 \
56	\| python3 "$PARSER" esearch
57
58	# Get article summaries as markdown table
59	bash "$EUTILS" fetch_json "16168343,16085191,31462531" \
60	\| python3 "$PARSER" esummary
61
62	# Get detailed metadata
63	bash "$EUTILS" fetch "16168343" \
64	\| python3 "$PARSER" efetch
65
66	# Generate BibTeX entries
67	bash "$EUTILS" fetch "16168343,16085191" \
68	\| python3 "$PARSER" bibtex
69
70	# Verify a citation by exact title
71	bash "$EUTILS" cite_lookup "Bivariate analysis of sensitivity and specificity" \
72	\| python3 "$PARSER" esearch
73
74	# Find related articles for a PMID
75	bash "$EUTILS" related "16168343" 10 \
76	\| python3 "$PARSER" esummary
77	```
78
79	Rate limiting: 3 requests/second without API key, 10/sec with NCBI_API_KEY.
80	The script auto-sleeps 350ms between calls. For batch operations, keep calls sequential.
81
82	E-utilities → MCP equivalence:
83
84	\| MCP Tool \| E-utilities Command \| Parser Mode \|
85	\|----------\|-------------------\|-------------\|
86	\| `search_articles` \| `search <query> [retmax]` \| `esearch` \|
87	\| `get_article_metadata` \| `fetch <pmids>` \| `efetch` or `bibtex` \|
88	\| `find_related_articles` \| `related <pmid> [retmax]` \| `esummary` \|
89	\| `lookup_article_by_citation` \| `cite_lookup <title>` \| `esearch` → `fetch` \|
90	\| `convert_article_ids` \| Not available (use CrossRef DOI lookup) \| — \|
91
92	---
93
94	## Workflow
95
96	### Phase 1: Search Strategy
97
98	1. Understand the need: Get the research topic, specific question, or manuscript section
99	that needs references.
100	2. Generate search terms:
101	- Identify key concepts (Population, Intervention/Exposure, Comparison, Outcome).
102	- Generate MeSH terms for PubMed queries.
103	- Build Boolean queries: `(concept1 OR synonym1) AND (concept2 OR synonym2)`.
104	3. Define scope:
105	- Date range (default: last 10 years unless user specifies).
106	- Article types (original research, review, meta-analysis, etc.).
107	- Language filter (default: English).
108	4. Present the search plan to the user before executing. Include the Boolean query,
109	databases to search, and filters.
110
111	*

People who liked this also viewed

6 picks

Type

search-lit

41installs53forks

View on GitHub

How to install search-lit?

aperivue/medsci-skills/search-lit

$npx -y skills add Aperivue/medsci-skills --skill search-lit

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Literature Search Skill
2
3	You are assisting a medical researcher with literature searches and citation management for
4	medical research papers. Every reference you produce must be verified against a live database --
5	never generate citations from memory alone.
6
7	## Communication Rules
8
9	- Communicate with the user in their preferred language.
10	- All citation content (titles, abstracts, BibTeX) in English.
11	- Medical terminology is always in English.
12
13	## Key Directories
14
15	- BibTeX output: User-specified directory (default: current working directory)
16	- Manuscript workspace: determined by the user or the calling skill
17
18	## Search Tools: MCP (Primary) + E-utilities (Fallback)
19
20	### Primary: MCP Tools (Claude.ai Remote)
21
22	\| Database \| MCP Tool \| Purpose \|
23	\|----------\|----------\|---------\|
24	\| PubMed \| `mcp__claude_ai_PubMed__search_articles` \| Search by query, MeSH terms \|
25	\| PubMed \| `mcp__claude_ai_PubMed__get_article_metadata` \| Full metadata for a PMID \|
26	\| PubMed \| `mcp__claude_ai_PubMed__find_related_articles` \| Related articles for a PMID \|
27	\| PubMed \| `mcp__claude_ai_PubMed__lookup_article_by_citation` \| Verify a citation \|
28	\| PubMed \| `mcp__claude_ai_PubMed__convert_article_ids` \| Convert between PMID/DOI/PMCID \|
29	\| Semantic Scholar \| `mcp__claude_ai_Scholar_Gateway__semanticSearch` \| Semantic search across all fields \|
30	\| bioRxiv/medRxiv \| `mcp__claude_ai_bioRxiv__search_preprints` \| Search preprint servers \|
31	\| bioRxiv/medRxiv \| `mcp__claude_ai_bioRxiv__get_preprint` \| Full preprint metadata \|
32	\| CrossRef \| WebFetch with `https://api.crossref.org/works/{DOI}` \| DOI verification \|
33
34	### Fallback: NCBI E-utilities (Direct API via Bash)
35
36	When PubMed MCP is unavailable (session timeout, "MCP session has been terminated" error,
37	or "No such tool available" error), fall back to NCBI E-utilities via bundled scripts.
38
39	Detection: If any `mcp__claude_ai_PubMed__*` call returns an error containing
40	"terminated", "not found", "not available", or "not connected", switch ALL subsequent
41	PubMed calls in this session to E-utilities. Do not retry MCP after a disconnect — it
42	will not recover within the same conversation.
43
44	Scripts (in `${CLAUDE_SKILL_DIR}/references/`):
45	- `pubmed_eutils.sh` — Bash wrapper for NCBI E-utilities API
46	- `parse_pubmed.py` — Python parser for E-utilities responses
47
48	Usage patterns:
49
50	```bash
51	EUTILS="${CLAUDE_SKILL_DIR}/references/pubmed_eutils.sh"
52	PARSER="${CLAUDE_SKILL_DIR}/references/parse_pubmed.py"
53
54	# Search PubMed (returns PMIDs)
55	bash "$EUTILS" search "diagnostic test accuracy meta-analysis radiology" 20 \
56	\| python3 "$PARSER" esearch
57
58	# Get article summaries as markdown table
59	bash "$EUTILS" fetch_json "16168343,16085191,31462531" \
60	\| python3 "$PARSER" esummary
61
62	# Get detailed metadata
63	bash "$EUTILS" fetch "16168343" \
64	\| python3 "$PARSER" efetch
65
66	# Generate BibTeX entries
67	bash "$EUTILS" fetch "16168343,16085191" \
68	\| python3 "$PARSER" bibtex
69
70	# Verify a citation by exact title
71	bash "$EUTILS" cite_lookup "Bivariate analysis of sensitivity and specificity" \
72	\| python3 "$PARSER" esearch
73
74	# Find related articles for a PMID
75	bash "$EUTILS" related "16168343" 10 \
76	\| python3 "$PARSER" esummary
77	```
78
79	Rate limiting: 3 requests/second without API key, 10/sec with NCBI_API_KEY.
80	The script auto-sleeps 350ms between calls. For batch operations, keep calls sequential.
81
82	E-utilities → MCP equivalence:
83
84	\| MCP Tool \| E-utilities Command \| Parser Mode \|
85	\|----------\|-------------------\|-------------\|
86	\| `search_articles` \| `search <query> [retmax]` \| `esearch` \|
87	\| `get_article_metadata` \| `fetch <pmids>` \| `efetch` or `bibtex` \|
88	\| `find_related_articles` \| `related <pmid> [retmax]` \| `esummary` \|
89	\| `lookup_article_by_citation` \| `cite_lookup <title>` \| `esearch` → `fetch` \|
90	\| `convert_article_ids` \| Not available (use CrossRef DOI lookup) \| — \|
91
92	---
93
94	## Workflow
95
96	### Phase 1: Search Strategy
97
98	1. Understand the need: Get the research topic, specific question, or manuscript section
99	that needs references.
100	2. Generate search terms:
101	- Identify key concepts (Population, Intervention/Exposure, Comparison, Outcome).
102	- Generate MeSH terms for PubMed queries.
103	- Build Boolean queries: `(concept1 OR synonym1) AND (concept2 OR synonym2)`.
104	3. Define scope:
105	- Date range (default: last 10 years unless user specifies).
106	- Article types (original research, review, meta-analysis, etc.).
107	- Language filter (default: English).
108	4. Present the search plan to the user before executing. Include the Boolean query,
109	databases to search, and filters.
110
111	*

People who liked this also viewed

6 picks

Type