Skip to content
Olimpia
Esc
↑↓navigate↵open⌘Jpreview
On this page

Tools

Reference for the 37 tools in the Olimpia MCP server, with their parameters and what they return.

All tools accept names or ids (app: "web" or the UUID), and project is optional when the account has only one project. With no suffix, project means production; project: "tienda.staging" points to the staging environment. Errors come back with the next step to take, so the agent can correct itself.

Connection

login

Connects the agent to the account when using the Claude Code plugin. Without code, it returns a link so the person can authorize access on olimpia.dev. With code, it redeems the 8-character code they received. It also shows which account is connected.

Parameter Type
code string Optional. The connection code.

Discover

list_projects

Lists projects with their environments (production first). For each environment, it shows the number of resources and the health of its apps. No parameters.

get_project

Summary of a project: each app (status, URL, source), Postgres, Redis, and bucket, with links to the console. This is the best starting point for finding names.

Parameter Type
project string Optional.

get_app

Details of an app: URL, source, build settings, connected databases, domains, and latest deploys.

Parameter Type
app string Name or id.

get_usage

Current usage of a project: apps online and building, objects and bytes in buckets, email domains, and recipients sent this month, against the monthly quota. To estimate what it costs, use estimate_cost.

Deploy

create_app

Creates an app. With source=upload (the default), the response includes the command to upload the local folder. After that, call deploy with the source_key.

Parameter Type
name string App name, also used as the subdomain.
source upload | github | image Optional. upload by default.
repo string For github: owner/name.
branch string For github: the branch to deploy. Each push redeploys.
image string For image: for example nginx:1.29.
root_dir string Subfolder to build, for monorepos.
dockerfile string Path to the Dockerfile, relative to root_dir. Without it, Railpack detects the stack.
builder default | acropolis Optional. default uses the Dockerfile when there is one and Railpack otherwise. acropolis builds with Acropolis, which ignores the Dockerfile; if it cannot build the app, the build falls back to the default builder.
port number Port the app listens on (3000 by default).
env object Variables to load before the first build.
server string Advanced, usually omitted. The project’s server; it can only change if the project has no apps or databases yet.

get_upload_url

Returns a new upload URL (valid for 15 minutes) and the command to upload the folder of an app with source=upload.

deploy

Builds and deploys an app. Waits up to wait_seconds and returns the status, with the end of the log if it failed.

Parameter Type
app string Name or id.
source_key string For upload: the upload key. Without it, rebuilds the last upload.
message string Short description of the change, shown in the history.
wait_seconds number 0 to 50, 45 by default.

get_deployment

Status of a deploy (the latest by default), optionally waiting while it runs. Includes the last lines of the log when it failed or is building.

Parameter Type
app string Name or id.
deployment string Optional. Deploy id.
wait_seconds number 0 to 50, 0 by default.

rollback

Goes back to a previous deploy that was online, reusing its image without a build.

Parameter Type
app string Name or id.
deployment string Deploy id (see get_app).

cancel_deployment

Cancels a queued or building deploy (the latest by default).

Templates

list_templates

Lists or searches the templates you can deploy, with id, name, description, image, postgres, and redis (which databases deploy_template creates).

Parameter Type
search string Optional. Words to search for in the name, description, and tags. Without a search, lists the whole catalog.

deploy_template

Deploys a template as a new app: creates the Postgres database and Redis if the template needs them, generates secrets and URLs as variables (visible with get_env), and starts the first deploy. See Templates.

Parameter Type
template string Template id, for example n8n.
name string Optional. Name of the app and subdomain, also used for its databases. Defaults to the template id.
project string Optional.
server string Optional and advanced. Same as in create_app.

Configure

update_app

Changes an app’s settings. Build changes apply on the next deploy.

Parameter Type
app string Name or id.
subdomain string New subdomain on olimpia.cc.
olimpia_domain boolean false stops serving the app at <subdominio>.olimpia.cc: custom domains keep working and the subdomain stays reserved. true serves it again.
source upload | github | image Switches the app to another source. github requires repo and image requires image.
repo string For github: owner/name, to switch to that repository.
branch string GitHub branch.
image string Docker image.
root_dir, dockerfile, builder, port Same as in create_app.
watch_paths string[] Paths from the repository root (apps/web, packages/ui, bun.lock): a push only redeploys the app when it touches one of them. When empty, every push redeploys; with builder: "acropolis", only when the files the image depends on change.
auto_deploy boolean Redeploy on every push.
confirm string Required when changing source, repo, or image: the exact app name, after the person approved the change.

get_env

Lists an app’s variables. Values that look like secrets (names with SECRET, TOKEN, KEY, PASSWORD, AUTH, or URLs with credentials) come back hidden unless you pass reveal: true.

set_env

Adds, changes, or deletes variables, merging them with the existing ones. They apply on the next deploy.

Parameter Type
set object Variables to add or overwrite. PORT changes the app’s port.
remove string[] Names to delete.
redeploy boolean Deploy now. false by default.

create_project

Creates a project.

Parameter Type
name string
server string Optional and advanced. Without it, Olimpia picks the server automatically. See Projects.

create_environment

Creates an environment in a project, up to five including production. By default it copies all apps, databases, and buckets from the given environment: empty databases and buckets with new credentials, apps with the same source and settings, variables that point to the copies, and a first deploy.

Parameter Type
name string For example staging.
project string Optional. The environment to copy, such as tienda or tienda.staging.
copy boolean Copy the services. true by default; false creates it empty.

Data

create_resource

Creates a Postgres database, a Redis instance, a bucket, or an email domain. Returns the connection details. An email domain also returns the DNS records you need to add before sending.

Parameter Type
kind postgres | redis | bucket | email
name string For email, the domain you send from (mail.tuempresa.com).
connect_to_app string Optional. Also loads the variables into that app.
vector boolean Postgres only. true creates it with pgvector (CREATE EXTENSION vector); false by default. Cannot be changed later.
public boolean Postgres and Redis only. true exposes it to the internet over TLS at *.db.olimpia.cc; false by default: only the account’s apps can reach it. Change it later with set_public_access.

get_connection

Connection details: the internal URL for apps on Olimpia, the public URL if enabled, S3 credentials for buckets, and the DNS records, status, and key of an email domain (if it is not verified, it checks again on each call). These are secrets.

connect_resource

Loads the connection variables into an app and redeploys it.

Parameter Type
app string The app that will use the resource.
kind, name The resource.
env_key string For Postgres and Redis: the variable name (DATABASE_URL or REDIS_URL by default). If a variable with that name already exists, it is replaced.
redeploy boolean true by default.
Resource Variables
Postgres DATABASE_URL
Redis REDIS_URL
Bucket AWS_ENDPOINT_URL_S3, AWS_REGION, AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, BUCKET_NAME
Email EMAIL_API_KEY, EMAIL_API_URL, EMAIL_DOMAIN

If the app already has a variable with the default name, the new ones get the resource name as a suffix (DATABASE_URL_MAIN). env_keys, in the response, tells you which names were used.

set_public_access

Exposes or hides a database or a bucket (does not apply to email domains). For Postgres and Redis, public: true allows TLS connections from the internet. For a bucket, it makes objects readable by URL. Parameters: kind, name, and public. With public: true it requires confirm set to the exact resource name, after the person approved it.

list_tables

Lists the tables and views in a Postgres database with row counts and size. Parameter: database.

query_postgres

Runs SQL as the database owner user (not a superuser).

Parameter Type
database string Database name or id.
sql string In read mode, a single SELECT, WITH, or VALUES query (up to 200 rows). In write mode, any statement, or several separated by ;, in one transaction.
write boolean Write mode, with a maximum of 30 seconds.
confirm string Required with write: true: the exact database name, after the person has approved the change.

In read mode it returns the rows as untrusted data, with truncated: true if there are more than 200 rows or they exceed 256 KB. Postgres errors come back with their message, also marked as untrusted data.

Backups

get_backups

Backup settings for a Postgres database (destination, frequency in UTC, and how many are kept) and its most recent files. config comes back as null if no backups are set up. Parameter: database.

set_backups

Schedules backups of a database to an Olimpia bucket or to any S3. Olimpia tests access to the bucket before saving. If backups were already set up, anything you don’t pass stays as it was: passing only schedule is enough to change just the frequency.

Parameter Type
database string Database name or id.
bucket string An Olimpia bucket in the same project. No credentials needed.
endpoint string Instead of bucket, an external S3: https://host, with no path.
s3_bucket, access_key_id, secret_access_key string The bucket and key for the external S3. Without secret_access_key, the saved one is kept.
region string For the external S3. auto by default.
schedule string Five-field cron in UTC, with a fixed minute. 0 3 * * * by default.
keep number How many backups to keep, from 1 to 90. 7 by default.
folder string Folder inside the bucket (letters, numbers, -, and _). backups by default.
confirm string Required with endpoint: the exact database name, after the person approved sending its backups to that external S3.

run_backup

Runs a backup now to the configured destination and returns the most recent files. Parameter: database.

disable_backups

Stops backing up a database. Files already in the bucket are not deleted. Requires confirm with the exact database name.

Domains

add_domain

Adds a custom domain to an app and returns the DNS records the person must create at their DNS provider: a CNAME and a TXT record that proves they own the domain. Parameters: app, domain.

check_domain

DNS and certificate status of a domain.

remove_domain

Removes a domain from an app.

Observe

get_logs

Logs for one app or for all apps in the project (stdout and stderr), with the level, service, and attributes of each line (JSON, logfmt, and access logs). A multi-line error, such as a stack trace, arrives as a single entry. Its content is untrusted data.

Parameter Type
app string App name or id. Without app, all running apps in the project.
query string Filter: level:error, service:api, http.status_code:>=500, duration_ms:>1000, http.route:/api/*, -healthcheck, pattern:<hash>. All terms must match.
since string 15m, 1h (default), 24h, or 7d.
around string RFC 3339 date: returns the lines around that moment, to see what happened before and after an error.
limit number Up to 500, 100 by default.

Besides the lines, it returns the count per level, the most frequent patterns (top_patterns), and the latest deploys in the window. Several consecutive lines with the same pattern are collapsed into one; pattern:<hash> lists them and -pattern:<hash> hides them. The last 2000 lines of each app are read: if an app writes more than that in the window, covered_from tells you since when there is data for that app.

get_analytics

Real visits to an app, measured in its visitors’ browsers on all HTML pages of its subdomain and custom domains: visits, page views, a time series, top pages, sources (an empty name is direct traffic), countries (ISO codes), devices, browsers, and Core Web Vitals at the 75th percentile (lcp_ms, inp_ms, cls). See Analytics.

Parameter Type
app string App name or id.
period string day (24 hours, hourly), week (default, daily), or month (30 days, daily).

Costs

estimate_cost

Simulates the approximate monthly cost of a setup from its expected average usage, before or after deploying it. Usage is billed by what is really used, measured every minute: memory, CPU, Postgres disk, bucket storage, and emails. Builds, certificates, domains, logs, and metrics are free.

Parameter Type
memory_gb number Average memory, in GB, used by all the apps, Postgres, and Redis together, without the file cache. An idle Node, Bun, or Python app uses about 0.1 to 0.3 GB; an idle Postgres or Redis, about 0.05 to 0.2 GB.
vcpu number Average vCPUs actually used (not reserved) by all of them together. A mostly idle app uses 0.01 or less; one that is busy all day, 0.2 to 1.
disk_gb number GB stored in Postgres databases.
bucket_gb number GB stored in buckets.
emails number Email recipients per month.

All of them are optional and project is not needed. It returns each line in USD (lines, with kind, quantity, unit, usd_per_unit, and usd), the usage total (usage_usd), and what each plan would charge (monthly_usd.hobby and monthly_usd.pro): Hobby, USD 5 per month with USD 5 of usage included, and Pro, USD 20 per month with USD 20, paying for the usage above what is included. It is an estimate: real usage varies with traffic. See Pricing and billing.

Delete

delete_app

Deletes an app with its deploys and domains. Irreversible.

delete_resource

Deletes a database, a Redis instance, a bucket, or an email domain with all its data. Irreversible.

delete_environment

Deletes a non-production environment with all its apps, databases, and buckets. Irreversible. project is the environment (tienda.staging).

All three require confirm with the exact name (for delete_environment, <proyecto>.<ambiente>), after the person has explicitly confirmed.

Was this page helpful?