Skip to main content
These eight tools cover discovery and read access. None of them mutate state, so you can safely chain them together to explore Alphacast.

​
search_catalog

Search the curated Alphacast catalog (default) or every public repository.
​
query
string
required
Search query. Tokenized server-side; multi-word queries score every term.
​
asset
string
default:"datasets"
Type of asset to return. One of datasets, charts, insights, repositories, accounts, dashboards, series, tickers.
​
limit
number
default:"10"
Maximum number of results.
​
offset
number
default:"0"
Pagination offset.
​
tags
string[]
Optional list of tag slugs to filter by.
​
repositoryId
number
Optional repository ID to scope the search to a single repo.
​
searchAllAlphacast
boolean
default:"false"
Set to true to search every public repository instead of just the curated catalog.
​
excludeDeprecated
boolean
default:"false"
When true, deprecated datasets are excluded from results.
Example prompt
“Find Argentina inflation series in the curated catalog.”
Underlying endpoint: GET /api/search.

​
get_dataset

Retrieve dataset metadata and the column schema.
​
datasetId
number
required
Numeric dataset ID. Strings are coerced.
Returns the dataset’s name, description, repository, last update timestamp, and the column array (name, dataType, isEntity). Example prompt
“Show me the schema of dataset 12345.”
Underlying endpoints: GET /datasets/{id} + GET /datasets/{id}/columns (called in parallel).

​
download_dataset

Download the current rows of a dataset.
​
datasetId
number
required
Numeric dataset ID.
​
format
string
default:"json"
json returns an array of objects. csv returns a CSV string.
Internally follows S3 redirects so the model receives the raw rows directly. Example prompt
“Download dataset 12345 as CSV and tell me the date range.”
Underlying endpoint: GET /datasets/{id}/data?$format=....

​
list_datasets

List every dataset the API key can read. No input parameters. Returns one row per dataset with id, name, repositoryId, and frequency. Example prompt
“List all datasets I have access to and pick the ones tagged ‘Argentina’.”
Underlying endpoint: GET /datasets.

​
list_repositories

List repositories where the API key can publish (Owner, Admin, or Write). No input parameters. Filtered server-side; read-only repositories are excluded so the model only sees repositories it can write to.
Tools that accept an optional repositoryId (for example create_pipeline) default to the user’s Home repository when no ID is provided — the server does not interrupt to ask you to pick one. Call list_repositories only when you want to write somewhere other than Home, or get_home_repository to check which repo Home resolves to.
Example prompt
“Which repos can I publish into?”
Underlying endpoint: GET /repositories (with client-side permission filter).

​
get_repository

Retrieve a single repository’s metadata.
​
repositoryId
number
required
Numeric repository ID.
Returns name, description, privacy, your permission level, and timestamps. Example prompt
“What’s the description of repository 1042 and who owns it?”
Underlying endpoint: GET /repositories/{id}.

​
get_home_repository

Resolve the user’s Home repository — the default workspace where pipelines and datasets land when no specific repository is named. Use it to inspect what Home points at before creating a pipeline, or to share its ID with collaborators. No input parameters. Returns the repository as if you had called get_repository on it. Example prompt
“What’s my Home repository?”
Underlying endpoint: GET /repositories/home.

​
get_dataset_profile

Comprehensive inspection of a dataset: column schema, entity values, date range and frequency, total row count, and a sample of rows. Use this for a quick understanding of what a dataset contains before downloading it.
​
datasetId
number
required
Numeric dataset ID.
​
sampleRows
number
default:"10"
Number of sample rows to include in the response. Maximum 50. Set to 0 to skip the sample entirely.
Returns dataset metadata, columns, entities, dateStats (date range and frequency), rowCount, and a sample array. Individual sections gracefully degrade if unavailable. Example prompt
“Profile dataset 12345 — how many rows does it have, what entities does it cover, and show me 5 sample rows.”
Underlying endpoints: GET /datasets/{id} + GET /datasets/{id}/columns + GET /datasets/{id}/entities + GET /datasets/{id}/date-stats + GET /datasets/{id}/data?$top=N (called in parallel).