Skip to main content
Glama

VibeDeploy

by be.vibedeploy
OverviewDiscussionsAdmin

Run container services (server, database, workers) for a site

deploy_app
Destructive

Run container services next to a site, from any public image: the user's own server (running code from the site's source tree), a database such as Postgres or MySQL, Redis, or workers, each in a gVisor sandbox. The call applies the whole config: services missing from it are stopped (their volumes are kept) and the others restart, picking up new source files. It waits until every service is ready and returns per-service status; if one is not ready, call get_app_status with logs. One service may be public: it serves the site's HTTP traffic on its port. Use "${generated:NAME}" for database passwords. API keys and other outside secrets never go in the config, the source or the chat: the user adds them in the VibeDeploy dashboard (site Settings, Secrets) and every service receives them as environment variables. Per site: up to 6 services, 2048Mi memory and 2000m CPU in total, 4 volumes, 3Gi storage. Container volumes are not included in snapshots yet.

Input Schema

NameRequiredDescriptionDefault
nameNoSite name as shown by list_sites, e.g. my-portfolio. Provide name or siteId.
configNoThe app config; omit it to use vibedeploy.json from the site's source tree. Shape: {"services":{"<name>":{"image":"postgres:16-alpine","port":5432,"public":false,"env":{"KEY":"value"},"volumes":{"/var/lib/postgresql/data":"1Gi"},"command":[],"args":[],"workingDir":"/app","source":false,"sourcePath":"/app","memory":"256Mi","cpu":"500m"}}}. Services reach each other by name (host "db"). "source":true copies the source tree into sourcePath on every start, e.g. {"image":"node:22-alpine","source":true,"command":["sh","-c","npm ci && node server.js"],"port":3000,"public":true}. "$(OTHER)" embeds an earlier env var of the same service, e.g. "DATABASE_URL":"postgres://app:$(DB_PASSWORD)@db:5432/app". For Postgres set PGDATA to a subdirectory of the volume.
siteIdNoSite id from list_sites or get_site. Use instead of name.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed3 schema fields changed
    • changedInput schema / properties / config / description
      Previous value: -"The app config object. Omit it to use vibedeploy.json from the site's source tree. App config: { \"services\": { \"<name>\": { \"image\": \"postgres:16-alpine\", \"port\": 5432, \"public\": false, \"command\": [...], \"args\": [...], \"workingDir\": \"/app\", \"env\": { \"KEY\": \"value\" }, \"volumes\": { \"/var/lib/postgresql/data\": \"1Gi\" }, \"source\": false, \"sourcePath\": \"/app\", \"memory\": \"256Mi\", \"cpu\": \"500m\" } } }. Any public image works. Services reach each other by service name (e.g. host \"db\", port 5432). Exactly one service may be \"public\": true; it receives all HTTP traffic for the site's domains on its \"port\" (plain HTTP, TLS is terminated for you). \"source\": true copies the site's source tree (write_source_files) into sourcePath (default /app, writable) on every start, so e.g. {\"image\":\"node:22-alpine\",\"source\":true,\"command\":[\"sh\",\"-c\",\"npm ci && node server.js\"],\"port\":3000,\"public\":true}. \"${generated:NAME}\" as an env value is a random secret generated once and shared by every service that references it (use it for database passwords); \"$(OTHER)\" embeds an earlier variable of the same service, e.g. \"DATABASE_URL\": \"postgres://app:$(DB_PASSWORD)@db:5432/app\". Secrets from outside (API keys, OAuth client secrets, tokens) never go in the config, the source or the chat: the user adds them in the VibeDeploy dashboard (the site's Settings tab, section Secrets; get_app_status returns the direct link as secrets.manageUrl). Every service gets them as env vars and restarts automatically; an explicit env entry with the same name wins. get_app_status lists the secret names that are set, never their values. Volumes persist across deploys; everything else is recreated. For Postgres set PGDATA to a subdirectory of the volume. Outbound internet works (not SMTP port 25); private networks are blocked. Limits per site: 6 services, 2048Mi memory and 2000m CPU in total (default 256Mi / 500m per service), 4 volumes, 3Gi storage."New value: +"The app config; omit it to use vibedeploy.json from the site's source tree. Shape: {\"services\":{\"<name>\":{\"image\":\"postgres:16-alpine\",\"port\":5432,\"public\":false,\"env\":{\"KEY\":\"value\"},\"volumes\":{\"/var/lib/postgresql/data\":\"1Gi\"},\"command\":[],\"args\":[],\"workingDir\":\"/app\",\"source\":false,\"sourcePath\":\"/app\",\"memory\":\"256Mi\",\"cpu\":\"500m\"}}}. Services reach each other by name (host \"db\"). \"source\":true copies the source tree into sourcePath on every start, e.g. {\"image\":\"node:22-alpine\",\"source\":true,\"command\":[\"sh\",\"-c\",\"npm ci && node server.js\"],\"port\":3000,\"public\":true}. \"$(OTHER)\" embeds an earlier env var of the same service, e.g. \"DATABASE_URL\":\"postgres://app:$(DB_PASSWORD)@db:5432/app\". For Postgres set PGDATA to a subdirectory of the volume."
    • addedInput schema / properties / name / description
      Added value: +"Site name as shown by list_sites, e.g. my-portfolio. Provide name or siteId."
    • addedInput schema / properties / siteId / description
      Added value: +"Site id from list_sites or get_site. Use instead of name."
  2. Changed1 schema field changed
    • changedInput schema / properties / config / description
      Previous value: -"The app config object. Omit it to use vibedeploy.json from the site's source tree. App config: { \"services\": { \"<name>\": { \"image\": \"postgres:16-alpine\", \"port\": 5432, \"public\": false, \"command\": [...], \"args\": [...], \"workingDir\": \"/app\", \"env\": { \"KEY\": \"value\" }, \"volumes\": { \"/var/lib/postgresql/data\": \"1Gi\" }, \"source\": false, \"sourcePath\": \"/app\", \"memory\": \"256Mi\", \"cpu\": \"500m\" } } }. Any public image works. Services reach each other by service name (e.g. host \"db\", port 5432). Exactly one service may be \"public\": true; it receives all HTTP traffic for the site's domains on its \"port\" (plain HTTP, TLS is terminated for you). \"source\": true copies the site's source tree (write_source_files) into sourcePath (default /app, writable) on every start, so e.g. {\"image\":\"node:22-alpine\",\"source\":true,\"command\":[\"sh\",\"-c\",\"npm ci && node server.js\"],\"port\":3000,\"public\":true}. \"${generated:NAME}\" as an env value is a random secret generated once and shared by every service that references it (use it for database passwords); \"$(OTHER)\" embeds an earlier variable of the same service, e.g. \"DATABASE_URL\": \"postgres://app:$(DB_PASSWORD)@db:5432/app\". Site secrets set in the dashboard are injected as env vars too. Volumes persist across deploys; everything else is recreated. For Postgres set PGDATA to a subdirectory of the volume. Outbound internet works (not SMTP port 25); private networks are blocked. Limits per site: 6 services, 2048Mi memory and 2000m CPU in total (default 256Mi / 500m per service), 4 volumes, 3Gi storage."New value: +"The app config object. Omit it to use vibedeploy.json from the site's source tree. App config: { \"services\": { \"<name>\": { \"image\": \"postgres:16-alpine\", \"port\": 5432, \"public\": false, \"command\": [...], \"args\": [...], \"workingDir\": \"/app\", \"env\": { \"KEY\": \"value\" }, \"volumes\": { \"/var/lib/postgresql/data\": \"1Gi\" }, \"source\": false, \"sourcePath\": \"/app\", \"memory\": \"256Mi\", \"cpu\": \"500m\" } } }. Any public image works. Services reach each other by service name (e.g. host \"db\", port 5432). Exactly one service may be \"public\": true; it receives all HTTP traffic for the site's domains on its \"port\" (plain HTTP, TLS is terminated for you). \"source\": true copies the site's source tree (write_source_files) into sourcePath (default /app, writable) on every start, so e.g. {\"image\":\"node:22-alpine\",\"source\":true,\"command\":[\"sh\",\"-c\",\"npm ci && node server.js\"],\"port\":3000,\"public\":true}. \"${generated:NAME}\" as an env value is a random secret generated once and shared by every service that references it (use it for database passwords); \"$(OTHER)\" embeds an earlier variable of the same service, e.g. \"DATABASE_URL\": \"postgres://app:$(DB_PASSWORD)@db:5432/app\". Secrets from outside (API keys, OAuth client secrets, tokens) never go in the config, the source or the chat: the user adds them in the VibeDeploy dashboard (the site's Settings tab, section Secrets; get_app_status returns the direct link as secrets.manageUrl). Every service gets them as env vars and restarts automatically; an explicit env entry with the same name wins. get_app_status lists the secret names that are set, never their values. Volumes persist across deploys; everything else is recreated. For Postgres set PGDATA to a subdirectory of the volume. Outbound internet works (not SMTP port 25); private networks are blocked. Limits per site: 6 services, 2048Mi memory and 2000m CPU in total (default 256Mi / 500m per service), 4 volumes, 3Gi storage."
  3. Added

TDQS

A4.6/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Rich disclosure well beyond the annotations: services missing from the config are stopped (volumes kept), the others restart and pick up new source files, the call blocks until every service is ready and returns per-service status, and it enumerates hard limits (6 services, 2048Mi/2000m, 4 volumes, 3Gi). It also discloses the secrets model (dashboard-added secrets become env vars, secrets must never go in config/source/chat) and the snapshot caveat, none of which the annotations convey.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

It is long but front-loaded — purpose and config-application behavior lead, operational details (public port, secrets, limits, snapshots) follow in order of relevance. Most sentences carry unique information; a few (e.g. the snapshot caveat) sit far from the main action but still earn their place.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a nested-object, high-consequence mutation with no output schema, the description supplies the return value (per-service status), failure routing, resource limits, and the secrets/volume-retention semantics. An agent has everything needed to invoke it correctly and interpret the result.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so the baseline is 3, but the description adds genuine meaning beyond the schema: the "${generated:NAME}" placeholder for DB passwords, the rule that API keys/secrets must come from the dashboard rather than config, and the overall config-application semantics. These are operational facts the schema's shape description does not establish.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description states a specific verb and resource — run container services (server, database, workers) next to a site — and immediately constrains the scope to gVisor-sandboxed containers from any public image. This clearly separates it from the many sibling deploy tools (deploy_site, build_and_deploy, begin_deploy), which an agent can distinguish without opening a schema.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

It gives a clear trigger (applying a whole services config next to a site) and routes the agent to an alternative on failure: 'if one is not ready, call get_app_status with logs.' However, it never explicitly contrasts this tool with other deploy siblings such as deploy_site or build_and_deploy, so the when-not-to-use boundary is left to inference.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All VibeDeploy toolsInspect
Try in Browser
Shareable, User-Owned Context BackendPeople, applications, and agents read, write, and coordinate together in one context lake. Reach it over MCP, a CLI, a Python SDK, or REST.fulcradynamics.comAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources