DOCSEARCHbyAlgolia
Submit Docs

Port Documentation

website·Indexed 36 days ago

https://docs.port.io/

Documentation for Port's software catalog and AI capabilities, including the AI Gateway, MCP server, and Context Lake. Covers defining ontologies with blueprints, managing AI agents and skills, implementing MCP connectors (such as GitHub), and configuring data access governance and aggregation properties.

Tokens
1.7M
Snippets
4.4K
Records
6.7K
Agent score
67%

What's inside Port

  1. Overview of Port AI

    Port AI is a foundational AI system and MCP (Model Context Protocol) client that enables natural language interaction with Port data. It accepts prompts, utilizes developer MCP tools from the Port MCP server to query the software catalog, and executes autonomous processes to return comprehensive responses. It is grounded in the Context Lake, meaning the accuracy of responses improves as the software catalog is more thoroughly modeled.

  2. Overview of Bitbucket (Self-Hosted) integration

    Port's Bitbucket (Self-Hosted) integration allows users to model and ingest data from Bitbucket Server and Bitbucket Data Center into a software catalog. It supports the following resources: Project, Repository, Pull Request, and User.

  4. Overview of Port Live Events

    Live events enable integrations to process source changes in real-time (typically via webhooks or watch APIs) to update mapped catalog entities between scheduled resyncs. This reduces the delay before changes appear in Port. Live events complement full and incremental syncs but do not replace full resync reconciliation. They continue to run even while a full resync is in progress.

  5. Overview of Port generic webhook ingestion

    Port's generic webhook integration allows you to ingest data into the software catalog from any service that provides outgoing webhooks, even without a native Port integration. It works by providing custom webhook endpoints that receive payloads from third-party services (e.g., GitHub, Sentry, PagerDuty). These payloads are processed using JQ JSON expressions to create, update, or delete entities in the software catalog.

  6. Overview of Port visualization widgets

    Port provides three categories of widgets to visualize software catalog data and add functionality to dashboards:

    1. Data Widgets: Visualization tools for catalog data, including Number charts, Pie charts, Bar charts, Line charts, Tables, and Entity cards.
    2. Custom Widgets: Functional widgets for extending the UI, including Iframes, Action cards, Action history, Markdown, AI Agents, Links, and Custom plugin UIs.
    3. Personal Widgets: User-specific widgets tailored to the logged-in user (available only on the Home page), including My entities, Recently viewed, and Recently used actions.

    Widgets can be placed on the Home page, within a dashboard tab on entity pages, or on standalone customizable dashboard pages in the software catalog.

  7. Overview of Port's Agentic SDLC Platform pillars

    Port provides a platform layer to run AI agents safely across the software development lifecycle (SDLC) based on six core pillars:

    • Context Lake: A unified engineering knowledge layer that powers AI reasoning, workflows, and dashboards.
    • Workflows & Tools: A visual, node-based automation builder for creating self-service experiences and complex operation chains triggered by users or catalog events.
    • Port AI: A conversational assistant that helps with setup, building workflows, and querying the catalog.
    • Agent Management: A governance layer for building AI agents, including an AI registry for managing skills, prompts, and MCP server access.
    • Interface Builder: A tool to create dashboards for real-time visibility into service health and deployment status, and to trigger on-demand workflows.
    • Governance: Manages access controls (RBAC, SSO, teams) and standards enforcement via scorecards for security, reliability, and production readiness.

  8. Overview of LLM provider options in Port

    Port offers two ways to power AI features (AI Assistant, AI Agents, etc.):

    1. Port's Managed AI Infrastructure (Default): Zero-setup, secured cloud infrastructure. Port does not train models on user data and ensures enterprise security with automatic updates.
    2. Bring Your Own LLM (BYOLLM): Allows organizations to route AI processing through their own infrastructure for stricter data privacy and regulatory compliance.

  9. Overview of Port's Datadog integration

    Port's Datadog integration allows users to model Datadog resources within Port and ingest data into them. Supported resources include Monitors, Services, SLOs, SLO History, Service Metrics, Users, Roles, Teams, Service Dependencies, Organizations, and Hosts.

  10. Overview of the Port Skills Registry

    A skills registry is a central library of reusable AI agent skills (prompts and instructions) that can be published and consumed across an organization. It prevents duplication of skills, ensures quality through governance bars (certification reviews, ROI scorecards), and provides visibility into usage and cost. Skills are discoverable via the Port UI, Port MCP, or CLI.

  11. Overview of GitHub Copilot metrics data model

    To visualize GitHub Copilot metrics in Port, a specific data model is required. This model uses a combination of integration-owned blueprints, foundation blueprints, and tool-agnostic AI profiles to aggregate daily activity into user, team, and organization-level KPIs.

    Blueprint Identifiers

    • GitHub Copilot User Usage (githubCopilotUserUsage): Daily per-user activity records.
    • GitHub Copilot Organization Usage (githubCopilotOrganizationUsage): Daily per-org activity records.
    • AI User Profile (ai_user): Per-tool AI activity profile (one per user per tool).
    • AI Org Profile (ai_org): Per-tool AI activity profile for the organization.
    • User (_user): Port users; surfaces individual Copilot metrics.
    • Team (_team): Port teams; rolls up member and child-team metrics.
    • Organization (organization): Holds organization-wide KPIs.

    Key Relations

    • AI Tools: Relation from User $\rightarrow$ AI User Profile. Links users to their AI profiles for rollups.
    • AI User Profile Relation: Relation from githubCopilotUserUsage $\rightarrow$ ai_user (identifier: copilot_user).
    • AI Org Profile Relation: Relation from githubCopilotOrganizationUsage $\rightarrow$ ai_org (identifier: copilot_org).

  12. Overview of GitLab-v2 Integration Improvements

    GitLab-v2 provides several enhancements over previous versions, including simplified authentication using a single service account token instead of multiple group tokens, and improved API efficiency for faster resyncs. It introduces typed configuration with first-class selectors for specific kinds such as merge-request, member, group-with-members, project-with-members, tag, release, and branch.