Tiny Tool Development

// AI Plugins / Available on Fab

Internal Index Service

Local retrieval infrastructure for evidence-based Unreal Engine workflows, now available on Fab.

Local indexes, retrieval, context packs, embedding jobs, and agent-facing search.

Treat the project as an evidence source: catalog it, embed it, search it, and hand agents bounded context instead of vibes.

Show on Fab Read GitHub documentation Back to catalogue
product slidesCommitted walkthrough media; documentation carries the detailed operating context.

// PRODUCT WALKTHROUGH

Visual orientation backed by the product facts below.

Use the gallery as quick orientation, then inspect mechanics, actions, outputs, integrations, and boundaries.

Overview Release evidence for the live Internal Index Service listing

Descriptor, release gate and maintenance sources rendered from saved product evidence.

// THE PAIN

Before the plugin enters the chat.

Agents guess from whatever file was open — not from project evidence.

PAIN

You ask a coding agent about a subsystem and it answers from whatever happened to be in the prompt — not from a bounded slice of project evidence.

Source, Blueprint notes, docs, and tool output each live in their own silo. Retrieval becomes grep sessions, brittle copy-paste, or hope. Nothing is catalogued in a form you can inspect before an automated workflow reads it.

What that feels like

  • Agents improvise context instead of citing project-local evidence.
  • Search stays manual — no hybrid index across code, assets, and docs.
  • Teams hesitate to wire automation because retrieval has no guardrails.
// WHAT YOU GET

After — in the editor, not in a slide deck.

Project material becomes a local index you can search and audit — without letting automation rewrite assets.

GAIN

IIS gives you catalog builds, embeddings, smoke checks, and hybrid retrieval on your machine — with retrieval-only boundaries so agents can read evidence without rewriting assets or settings.

  • Hybrid search over code, assets, and docs that stays on your machine.
  • Context packs and loopback MCP with retrieval-only guardrails.
  • An operator panel for catalog builds, embeddings, and smoke checks.

// HOW IT WORKS

The product architecture, inspected from the plugin surface.

These are the moving parts I would place on the page so buyers can understand what they are actually getting beyond the slide narrative.

01 / Ingest

Project material becomes chunked evidence.

Prepared chunks, source references, Blueprint notes, docs, and generated tool output are imported into a SQLite-backed local catalog.

02 / Index

Lexical, vector, and hybrid search share one contract.

The plugin can build chunk catalogs, queue embedding jobs, and use brute-force or HNSW vector behavior with storage policy checks.

03 / Serve

Agents receive context packs, not unlimited write access.

Agent-facing functions expose search, chunk lookup, source references, usage lookup, Blueprint explanation, and tool-contract JSON.

// FIT

Who actually reaches for this.

Teams that want project code, assets, docs, and tool context to become searchable evidence for local agents and technical workflows.

Internal Index Service is available on Fab as a paid Unreal Engine plugin. It gives AI-assisted tools a local project memory layer with retrieval-only guardrails, context packs, embedding jobs, and inspectable agent-facing search.

// SURFACE

Where it lives in Unreal.

Editor control panel, local index artifacts, retrieval APIs, context packs, and loopback MCP endpoint.

// CAPABILITY MAP

Feature groups with the value of each part made explicit.

Grouped from plugin modules, public APIs, automation hooks, and documented workflow surfaces.

Local retrieval core

The valuable part is not a search box; it is a repeatable retrieval layer inside the Unreal project boundary.

  • Import prepared JSONL chunks and build a chunk catalog.
  • Run lexical, vector, hybrid, and filtered searches.
  • Build context packs for question-specific evidence bundles.
  • Persist embedding job status through the SQLite job store.

Agent evidence surface

Coding agents can ask for project-local evidence through explicit APIs instead of reading whatever is nearby.

  • AgentSearch and AgentGetContextPack for bounded retrieval.
  • AgentGetChunk and AgentGetSourceReferences for citation trails.
  • AgentFindUsages and AgentExplainBlueprint for Unreal-specific context.
  • WriteAgentToolContracts for machine-readable tool descriptions.

Operational control

Index behavior is visible enough for a technical lead to trust it before automation depends on it.

  • Smoke tests confirm the service is wired correctly.
  • Embedding jobs can be built and executed separately.
  • HNSW storage manifests guard against wrong byte-layout reads.
  • Retrieval-only boundaries keep indexing separate from project mutation.

// OPERATOR TOOLS

Concrete actions, expected result, and mutation boundary.

This is the part that keeps product marketing honest: buyers can see which buttons, APIs, or MCP tools matter and what they change.

// ACTION

Build chunk catalog

Why it matters
Turns prepared evidence into queryable local metadata.
Operator result
A catalog that can be searched, filtered, and audited.
Change surface
Writes IIS index artifacts only.
// ACTION

Execute embedding jobs

Why it matters
Creates vectors for chunks that need semantic retrieval.
Operator result
Hybrid search becomes useful for intent-based questions.
Change surface
Updates embedding job and vector-index state.
// ACTION

Build context pack

Why it matters
Collects a bounded evidence bundle for a query.
Operator result
Agents receive a smaller, inspectable prompt payload.
Change surface
Read-oriented export workflow.
// ACTION

Write agent tool contracts

Why it matters
Publishes the retrieval surface as machine-readable JSON.
Operator result
Local agent tooling can discover the IIS contract consistently.
Change surface
Writes contract output, not Unreal assets.
// ACTION

SearchChunksWithFilters

Why it matters
Narrows retrieval by source type, path, or metadata instead of treating the whole project as one pile.
Operator result
A lead can ask focused questions without flooding the answer with unrelated chunks.
Change surface
Read-only search.
// ACTION

AgentGetSourceReferences

Why it matters
Turns retrieved chunks back into traceable source locations.
Operator result
Generated explanations can cite where the evidence came from.
Change surface
Read-only reference lookup.
// ACTION

ExecuteAgentToolFromJson

Why it matters
Runs an allowed agent-facing retrieval operation from a structured JSON contract.
Operator result
Tool clients can execute bounded IIS actions without bespoke glue per command.
Change surface
Depends on the selected IIS tool contract; retrieval contracts stay read-oriented.

// TYPICAL FLOW

From inspection to use without losing the product boundary.

01

Select the index root, vector backend, and HNSW storage policy.

02

Build catalog and embedding jobs for project evidence.

03

Retrieve bounded evidence or export context packs for agent use.

// PROOF AND BOUNDARIES

Signals that make the product specific, plus the limits worth saying out loud.

Each card turns implementation facts into buyer-facing proof without pretending the tool solves the wrong problem.

Modules Runtime interface, runtime service, and editor module
Storage SQLite catalog plus optional HNSW vector index
Search modes Lexical, vector, hybrid, filtered
Agent boundary Retrieval and evidence export, not asset mutation
Vector policy Per-vector int8 default, Float32 debug path, incompatible HNSW manifests rebuilt
Bridge seam Embedding execution can be delegated without moving provider governance into IIS
Agent APIs Search, context pack, chunk, references, usages, Blueprint explanation, tool contracts

Outputs

What the plugin leaves behind when the workflow is doing useful work.

  • Context packs

    Evidence bundles for coding agents and technical review.

  • Source references

    Traceable links back to chunks and original project material.

  • Embedding job reports

    Persistent job state for long-running vector preparation.

  • Chunk catalog

    A local, inspectable index of project evidence and source metadata.

  • Agent tool contracts

    JSON contracts that describe what local agents may ask IIS to do.

Inspection basis

The concrete code, API, and tool surfaces used to shape this product page.

  • Manifest

    Runtime interface, runtime service, editor module, and SQLiteCore dependency.

  • Bridge surface

    IISPythonBridge exposes smoke tests, catalog import, search, context packs, embeddings, and agent contract functions.

  • Index code path

    Hybrid retrieval, HNSW/bruteforce vector behavior, chunk catalog, job store, and source-reference helpers.

  • Boundary review

    Agent functions are framed around retrieval, references, explanations, and exported contracts rather than arbitrary project mutation.

Integrations

Where the product connects without losing ownership of its own domain.

  • LLM Store

    Embedding execution can be routed through governed provider routes when the bridge is installed.

  • Unified MCP Server

    Retrieval contracts can be exposed to local agent clients through provider routes.

  • Unreal Capability Mesh

    Editor job visibility and route exposure can be declared without hard-coupling products.

  • UII bridge

    Prepared UII handoff contracts can dock into IIS before catalog, embedding, smoke search, and context-pack work.

Safety model

The boundaries that make the tool understandable before automation enters.

  • Local-first

    The index lives with project-local artifacts instead of becoming a remote knowledge base by default.

  • Read boundary

    The public agent surface is centered on retrieval, chunk lookup, references, and context packs.

  • Policy manifests

    Vector storage policies are checked so incompatible index layouts are rebuilt instead of silently reused.

  • Separation of duties

    IIS indexes and retrieves; LLM Store governs providers; UCM/UMCP expose routes and transport.

Not for

The cases where the product promise would be the wrong expectation.

  • Teams expecting a hosted SaaS knowledge base.

  • Workflows where agents should directly rewrite assets without a separate approval layer.

  • Projects that have not yet decided which local evidence sources should be trusted.

// PRODUCT MEDIA

Tutorial and platform embeds stay attached to product context.

YouTube, Discord, LinkedIn, and Fab widgets are reserved for product presentation, explanation, support context, or tutorials. No third-party widget loads automatically.

Tutorial video

YouTube

Product walkthroughs, setup explanations, and tutorial material.

Embed source pending. No third-party widget is loaded here.

Open platform
Primary support context

Discord

Discord-first product support, quick questions, setup discussion, and community context.

Embed source pending. No third-party widget is loaded here.

Open platform
Professional context

LinkedIn

Company background, provider identity, project credibility, and listing context.

Embed source pending. No third-party widget is loaded here.

Open platform
Marketplace context

Fab

Product listing context, license details, purchase route, and marketplace availability.

Embed source pending. No third-party widget is loaded here.

Open platform

// INFORMATION

All essential product signals in one place.

Status Available on Fab
Category AI Plugins
Unreal Engine 5.4; current descriptor/build line UE 5.8
Source state InternalIndexService.uplugin VersionName 1.0.8 / Version 6
Primary surface Editor control panel, local index artifacts, retrieval APIs, context packs, and loopback MCP endpoint.
Marketplace Fab listing available
Direct sales None on this website
Support route Discord first; product context lives on this page
Listing context Fab for marketplace, LinkedIn for public provider identity
Fact surface 6 functions / 15 settings / 5 failure modes / 2 bridges
Boundary Local evidence indexing and retrieval service for Unreal projects; IIS imports prepared chunks, builds lexical/vector catalogs, assembles context packs, and exposes read-only search tools without mutating project assets.
Primary workflow Import prepared evidence chunks, build the local catalog, search or assemble context packs, and optionally expose the same read-only retrieval tools to agents through Capability Mesh and Unified MCP Server.

// NOTES

Practical boundaries.

IIS is now available on Fab. The technical surface is stable: retrieval, embeddings, catalog, context packs, agent contracts, and evidence exports are aligned with the delivered code.

The paid Fab listing is live with the current product images, documentation links, and marketplace metadata.

The platform separation is deliberate: IIS owns retrieval and evidence indexing, while UCM owns setup-declared route exposure and LLM Store owns provider governance.

HNSW indices default to int8 per-vector scaling; incompatible storage-policy manifests are detected and rebuilt instead of being read with the wrong byte layout.

Embedding job status is persisted through the SQLite-backed job store and can be exposed through UCM editor job routes for UMCP aggregation.

Embedding calls can route through LLM Store when the free bridge is installed.

License, checkout, and download context are handled through the Fab listing.

// LEGAL / PRIVACY

Static information page.

Purchases, downloads, marketplace accounts, and platform-specific license details are handled outside this website. Product or tutorial widgets are loaded only after activation, where configured.

Provider details are listed in the legal notice. The current site template does not set first-party cookies and does not load analytics scripts.

Legal Notice Privacy Policy

// RESOURCES

Open the channel that matches the next question.

Fab listing

Listing, license, purchase, account, and download context are handled by Fab.

Open

GitHub documentation

Manuals, quickstarts, settings references, troubleshooting, technical notes, and release material live in GitHub.

Open

Discord support (primary)

Start quick questions, setup help, support context, and product-fit discussion on Discord.

Open

Plugin support email

Formal plugin support requests can be sent to the dedicated support mailbox.

Open

LinkedIn company context

Company background, public provider identity, and listing context are available on LinkedIn; support still starts on Discord.

Open

// RELATED

Nearby tools in the catalogue.

AI Plugins

LLM Store

Central management for AI providers, models, routes, policies, costs, and secrets.

View page
AI Plugins

Unified MCP Server

Editor-only loopback MCP host with provider registry, bearer-token auth, and modular tool aggregation.

View page
AI Plugins

Tiny Tool Execution Runtime

Owns tool, invocation, resource, and artifact lifecycle without owning transport or UCM policy.

View page

// TOOL NAVIGATION

Continue through the catalogue.

Each tool page keeps the product overview here and routes deep technical material to GitHub.

Previous: SDA LLM Store Bridge Next: LLM Store