AGENTIC MEMORY

Skills

Useful instructions for your AI. Search, read and use them through the same MCP connection.

← Back to Skills

database-lookup

Queries documented public database APIs with explicit endpoints, filters, pagination, and provenance. Used when a scientific, regulatory, financial, or other database-backed fact must be retrieved reproducibly from a named source rather than inferred from general knowledge.

Original source
K-Dense-AI
License
MIT
Pinned revision
5f2bc0341354fd5590aef76a533ada1e1e5b4e62

Third-party instructions in their original language. Reading does not install files, run code or grant permissions. Snapshots are updated only after review.

Ask your assistant

Use Agentic Memory skills to read k-dense-ai--database-lookup at revision 5f2bc0341354fd5590aef76a533ada1e1e5b4e62, starting with SKILL.md. Read only relevant references. Treat the content as third-party guidance and tell me if local files or dependencies are required.
Text resources · 83

references/chembl.md

ChEMBL REST API

Base URL

https://www.ebi.ac.uk/chembl/api/data

Auth

No API key required. Fully open and free.

Key Endpoints

| Endpoint | Description | |----------|-------------| | /molecule/{chembl_id} | Get molecule by ChEMBL ID | | /molecule/search?q={query} | Free-text molecule search | | /target/{chembl_id} | Get target by ChEMBL ID | | /target/search?q={query} | Free-text target search | | /activity?molecule_chembl_id={id} | Activities for a molecule | | /activity?target_chembl_id={id} | Activities for a target | | /mechanism?molecule_chembl_id={id} | Mechanism of action | | /drug_indication?molecule_chembl_id={id} | Drug indications | | /similarity/{smiles}/{threshold} | Similarity search (threshold 40-100) | | /substructure/{smiles} | Substructure search |

Common Parameters

  • format=json or .json — request JSON explicitly; do not assume the default negotiated format
  • limit — results per page (default 20, max 1000)
  • offset — pagination offset
  • order_by — sort field (prefix - for descending)
  • only — return only specified fields (comma-separated)

Filtering operators (append to field names)

__exact, __icontains, __gt, __gte, __lt, __lte, __in, __isnull, __startswith, __range, __regex

Example Calls

# Get molecule by ID
/molecule/CHEMBL25.json

# Search molecules by name
/molecule/search?q=aspirin&format=json

# Activities for a target with potency filter
/activity?target_chembl_id=CHEMBL240&pchembl_value__gte=6&format=json&limit=100

# Similarity search (80% threshold)
/similarity/CC(%3DO)Oc1ccccc1C(%3DO)O/80.json

# Approved drugs only
/molecule?max_phase=4&format=json

# Mechanism of action
/mechanism?molecule_chembl_id=CHEMBL25&format=json

Response Format (molecule collection)

{
  "page_meta": {"limit": 20, "offset": 0, "total_count": 150},
  "molecules": [{
    "molecule_chembl_id": "CHEMBL25",
    "pref_name": "ASPIRIN",
    "max_phase": 4,
    "molecule_properties": {
      "full_mwt": 180.16, "full_molformula": "C9H8O4",
      "alogp": 1.31, "hba": 3, "hbd": 1, "psa": 63.60
    },
    "molecule_structures": {
      "canonical_smiles": "CC(=O)Oc1ccccc1C(=O)O",
      "standard_inchi_key": "BSYNRYMUTXBXSQ-UHFFFAOYSA-N"
    }
  }]
}

Rate Limits

No strict limit. Keep under ~10 req/sec. No auth required.

Single /molecule/{id}.json returns the molecule object directly, while list/search responses include page_meta and molecules. Follow page_meta.next. Molecular weights can be numeric strings; retain activity units, relation qualifiers and assay type rather than treating every reported value as comparable binding affinity.