AWS Agent Registry adds custom metadata schemas for governed agent discovery

· Engineer's Notes · Cem Koyluoglu

Custom metadata schemas enforce organization‑wide attributes on Agent Registry records, enabling consistent governance, compliance checks, and searchable discovery.

What happened

AWS Agent Registry introduced custom metadata schemas that allow an administrator to define a set of organization‑defined key‑value pairs. Every record—such as agents, Model Context Protocol (MCP) servers, or skills—must adhere to the schema when created or updated. The schema specifies field names, data types (Text, Enum, URL, Boolean), allowed values, and required versus optional status. In the AnyCompany example, the default schema requires environment, businessUnit, and securityScanStatus, while optional fields include ownerTeam, cmdbId, dataClassification, and securityScanReportUrl. An MCP‑specific override adds an optional networkEgress field. Records submitted without required metadata are stored as NON_COMPLIANT and cannot be approved. Invalid values trigger a ValidationException. The registry validates metadata on each write, stores compliance status, and emits events to Amazon EventBridge when a record moves to Pending approval. A Lambda function can then read the metadata, verify compliance (e.g., securityScanStatus = PASSED, presence of cmdbId), and approve or reject the record. Approved records become searchable via discovery APIs that support filtering on custom attributes such as environment=prod AND businessUnit=payments.

Why it matters in production

This capability standardizes attribute vocabularies across teams, reducing the risk of divergent naming or value conventions that can break automated governance pipelines. By enforcing schema compliance at write time, organizations gain early detection of mis‑configurations, lowering downstream remediation costs. The compliance flag (COMPLIANT vs NON_COMPLIANT) provides a deterministic gate before approval, enabling automated security and cost‑center checks without manual review. Indexed custom metadata improves query latency for discovery operations, as filters operate on pre‑validated fields rather than full‑text scans. Integration with EventBridge and Lambda allows real‑time enforcement of security policies (e.g., rejecting records with securityScanStatus = FAILED), which strengthens the security posture. Finally, the ability to attach CMDB IDs and cost‑center data creates a single source of truth for asset inventory, facilitating cost allocation and audit reporting.

Engineering takeaways

  • Define a minimal required schema (e.g., environment, businessUnit, securityScanStatus) to guarantee essential governance data while keeping onboarding friction low.
  • Implement automated validation pipelines that capture ValidationException responses and correct enumerated values before retrying record creation.
  • Use EventBridge to trigger Lambda approval functions that read custom metadata, enforce compliance rules, and route non‑compliant or high‑risk records to manual review.
  • Index custom metadata fields and design discovery queries that combine natural‑language terms with explicit filters to achieve fast, scoped results.
  • Align schema definitions with existing CMDB, classification, and scanning tools so that automation can populate optional fields (cmdbId, dataClassification, securityScanReportUrl) before submission.

Related work on this site

  • Agentic & LLM Automation — Multi-step LLM pipelines that run unattended: model cascades with fallbacks, validation gates, scheduled automation and alerting when something breaks.
  • RAG & Grounded LLM Systems — LLM features that answer from the right source instead of from memory: document-grounded assistants, transcript-grounded chat and quality gates around generated output.
  • YouTube AI Summarizer — Open-source Chrome extension on the Chrome Web Store: AI summaries, key points, deep analysis, a two-host AI podcast (Gemini TTS) and transcript-grounded chat for any YouTube video, using Groq or Ollama Cloud with the user’s own key.
  • Automated AI News Pipeline (this site) — The pipeline behind this site’s Tech News: scheduled GitHub Actions scrape sources, an LLM cascade on Groq rewrites and enhances articles, and layered quality gates (date integrity, language checks, instruction-leak detection, duplicate detection) decide what gets published, with Telegram alerts.

This note was drafted with AI assistance from the primary source credited on this page, and automatically checked against that source before publishing.

Source: https://aws.amazon.com/blogs/machine-learning/govern-and-discover-records-with-common-metadata-in-aws-agent-registry/

← All tech news