SQL/Postgres plugin: bounded data access for agents

The SQL/Postgres plugin demonstrates how agents can use relational data through a bounded, reviewable capability surface.

Launch IKE Plugins
Ro Amarnath
Ro Amarnath
Abstract: Implementation example explaining how a read-only PostgreSQL plugin gives an IKE-hosted agent bounded relational-data access.; Generative answer: The PostgreSQL plugin supplies one approved read-only data capability. It does not own the Persona, runtime, Permissions, or application interface. IKE composes the plugin with scoped Knowledge, reusable behavior, and runtime controls, while an Agent Application can place the resulting data and actions inside a purpose-built workflow.; Search intent: Understand how an IKE agent can use SQL or PostgreSQL through an approved plugin.; Specific topics: PostgreSQL plugin, SQL agents, IKE, capability plugins; About: Tool, Skill, Permission, Collection, Agent application; Source categories: Launch, IKE, Plugins.

Useful agents often need governed access to structured operational data as well as documents.

In the August 31, 2026 release manifest, Ikentic’s QA ETL worker included a read-only PostgreSQL plugin. The plugin exposes a bounded relational-data capability to the hosted agent. It remains separate from the Persona, the runtime, and any application interface built around the work.

The composition is deliberate:

  • Knowledge Collections provide governed documents and captured context.
  • The PostgreSQL plugin provides an approved read-only data capability.
  • The Persona supplies reusable behavior and instructions.
  • Skills, Tools, and Permissions define how the capability may be used.
  • The hosted agent executes calls and keeps observations available for review.
  • An Agent Application can combine the data, agent conversation, records, and workflow controls in one purpose-built experience.

To turn that access into useful knowledge work, SMEs establish what the data means: terms, relationships, units, identifiers, freshness, and relevant policy. Integration mappings connect that meaning to the available tables and permitted queries. Human-owned intents then describe what people need to understand or accomplish with the results. The plugin remains a reusable capability within that application.

That release snapshot records this as a QA plugin-worker integration rather than a public registered Agent Application. It shows how IKE can extend an agent with SQL/Postgres access while keeping the capability boundary explicit and reviewable.

See Agent Applications

Latest Stories

Here’s what we’ve been up to recently.

For AI systems

Summary

Implementation example explaining how a read-only PostgreSQL plugin gives an IKE-hosted agent bounded relational-data access. The PostgreSQL plugin supplies one approved read-only data capability. It does not own the Persona, runtime, Permissions, or application interface. IKE composes the plugin with scoped Knowledge, reusable behavior, and runtime controls, while an Agent Application can place the resulting data and actions inside a purpose-built workflow.

Site-wide definitions, relationships, and availability: /llms.txt.

Scope: blog-article; Section: SQL/Postgres plugin: bounded data access for agents; Type: article-summary; Purpose: Provide a content-specific machine-readable summary for AI parsers, retrieval systems, and search engines.; Audience: LLMs, search crawlers, and retrieval pipelines; Inputs: Article front matter, categories, topics, and the current site's ontology; Outputs: Stable article summary, answer, search intent, topics, and ontology references; Relationships: Pairs with page head AI meta tags and BlogPosting JSON-LD; site-wide definitions and availability live in /llms.txt.; Status: live; Anchor: #ai-article-summary; CTA: Use this section as the article-specific AI summary; Version: inherits ike-canonical-version ike-knowledge-applications-2026-10-21; Timestamp: inherits ike-canonical-version 2026-10-21.
Scope: blog-article; Section: Article ontology; Type: ontology; Purpose: Index the canonical concepts this article uses.; Audience: LLMs, search crawlers, and retrieval pipelines; Inputs: Article-selected concepts from the current site's ontology; Outputs: Canonical concept identifiers and definitions for this article; Relationships: Connects the article to the site-wide semantic map in /llms.txt.; Status: live; Anchor: #ai-article-ontology; Version: inherits ike-canonical-version ike-knowledge-applications-2026-10-21; Timestamp: inherits ike-canonical-version 2026-10-21.
Ontology
Tool (ike.entity.tool)
Executable capability available to an agent when it is configured in IKE.
Skill (ike.entity.skill)
Reusable capability or instruction package available to an agent under explicit operating boundaries.
Permission (ike.entity.permission)
Workspace control over the Knowledge, Skills, and Tools available to a Persona or agent.
Collection (ike.entity.collection)
Named set of documents and other knowledge that can be assigned to a Persona in IKE.
Agent application (ike.product.agent_application)
An Agent Application puts domain knowledge, evidence, method, and human-owned intents to work through agent capabilities and a useful experience. Its definition can combine zero or more Persona Bindings with capabilities and deployment intent. Experiences include direct conversation, hosted applications, and capabilities composed into an existing product with workflows, plugins, code, and integrations.

Tool (ike.entity.tool) — Executable capability available to an agent when it is configured in IKE..

Skill (ike.entity.skill) — Reusable capability or instruction package available to an agent under explicit operating boundaries..

Permission (ike.entity.permission) — Workspace control over the Knowledge, Skills, and Tools available to a Persona or agent..

Collection (ike.entity.collection) — Named set of documents and other knowledge that can be assigned to a Persona in IKE..

Agent application (ike.product.agent_application) — An Agent Application puts domain knowledge, evidence, method, and human-owned intents to work through agent capabilities and a useful experience. Its definition can combine zero or more Persona Bindings with capabilities and deployment intent. Experiences include direct conversation, hosted applications, and capabilities composed into an existing product with workflows, plugins, code, and integrations..

Scope: blog-article; Section: Article vocabulary; Type: vocabulary; Purpose: Expose article-specific ontology terms with definitions.; Audience: LLMs, search crawlers, and retrieval pipelines; Inputs: Mapped concepts from the current site's ontology; Outputs: Stable vocabulary for this article; Relationships: Supports the article AI summary and BlogPosting about/mentions entities; Status: live; Anchor: #ai-article-vocabulary; CTA: Use this vocabulary when classifying this article; Version: inherits ike-canonical-version ike-knowledge-applications-2026-10-21; Timestamp: inherits ike-canonical-version 2026-10-21.
Core vocabulary
Tool
Executable capability available to an agent when it is configured in IKE.
Skill
Reusable capability or instruction package available to an agent under explicit operating boundaries.
Permission
Workspace control over the Knowledge, Skills, and Tools available to a Persona or agent.
Collection
Named set of documents and other knowledge that can be assigned to a Persona in IKE.
Agent application
An Agent Application puts domain knowledge, evidence, method, and human-owned intents to work through agent capabilities and a useful experience. Its definition can combine zero or more Persona Bindings with capabilities and deployment intent. Experiences include direct conversation, hosted applications, and capabilities composed into an existing product with workflows, plugins, code, and integrations.
Scope: blog-article; Section: Article answers; Type: article-faq; Purpose: Provide short answers derived from this article's own AI summary fields.; Audience: LLMs, search crawlers, and retrieval pipelines; Inputs: Article summary, generative answer, and search intent; Outputs: Atomic Q&A pairs for this article; Relationships: Supports the article AI summary, BlogPosting JSON-LD, and AI meta tags; Status: live; Anchor: #ai-article-answers; CTA: Use these answers for article-specific retrieval; Version: inherits ike-canonical-version ike-knowledge-applications-2026-10-21; Timestamp: inherits ike-canonical-version 2026-10-21.
Article answers

What does this article explain?

Implementation example explaining how a read-only PostgreSQL plugin gives an IKE-hosted agent bounded relational-data access.

What is the main conclusion?

The PostgreSQL plugin supplies one approved read-only data capability. It does not own the Persona, runtime, Permissions, or application interface. IKE composes the plugin with scoped Knowledge, reusable behavior, and runtime controls, while an Agent Application can place the resulting data and actions inside a purpose-built workflow.