Helix MCP Knowledge

Open source · Public release

Evidence-grounded knowledge for BMC Helix

Find the evidence.
Keep the provenance.

A local MCP server that turns selected official documentation and authorised project knowledge into retrievable, source-linked context for AI agents.

Explore the retrieval model View source
KNOWLEDGE SEARCH / 01LOCAL INDEX
Query

How is access controlled?

↵
ScopeOfficial docsSelected project
010.94

Official documentation

Access control and permissions

The matching section is returned with its source URL, document identity and chunk reference.

canonical URLdocument IDchunk ID

Evidence returned with enough context to verify.

What it solves

Large documentation spaces, made usable.

Knowledge gives agents a documentary surface that remains explicit about source, scope and project context.

01

Search at useful scale

Retrieve relevant sections across selected BMC documentation without manually navigating a large documentation estate.

02

Ground the context

Return canonical URLs and stable document and chunk identifiers alongside the matching text.

03

Keep projects isolated

Combine official sources with explicitly selected project documentation without blending separate knowledge spaces.

04

Operate the lifecycle

Use a local dashboard to select, synchronise, maintain and remove documentation intentionally.

Knowledge model

Two source classes.
One explicit scope.

Official product documentation and authorised project files enter separate, traceable ingestion paths. The effective search scope is resolved before retrieval.

A
Official documentationSelected products and versions
B
Project documentationExplicitly authorised local files
C
Local knowledge indexText, metadata, provenance and optional vectors
D
Evidence resultRelevant section with source identity

Retrieval model

Useful from the baseline.
Extensible when needed.

The default local index supports evidence retrieval without optional model components. Additional capabilities can be enabled and removed deliberately.

COREIncluded

Lexical retrieval

Fast text search, explicit source filters, project resolution and section expansion from the local index.

  • Keyword and phrase matching
  • Scope-aware ranking
  • Neighbouring section expansion
OPTIONEnable locally

Scanned PDF OCR

Extract searchable text from authorised scanned project PDFs when the source requires it.

OPTIONEnable locally

Semantic search

Add vector-based retrieval for meaning-aware matches across the selected knowledge scope.

OPTIONEnable locally

Result reranking

Refine candidate ordering when a more precise relevance pass is useful for the workflow.

Local lifecycle

From source selection to verifiable context.

  1. 01

    Select

    Choose official product versions and define authorised project knowledge spaces.

  2. 02

    Synchronise

    Download or scan only the selected sources and persist their identity locally.

  3. 03

    Retrieve

    Search an explicit scope and expand the sections needed for useful context.

  4. 04

    Verify

    Carry provenance into the agent response so the underlying evidence can be inspected.

Product boundary

Knowledge retrieves documentary evidence. It does not connect to live Helix environments or perform operational changes.
For controlled action, explore Helix MCP Gateway

Release status

Now open source.

Helix MCP Knowledge is publicly available for review, installation and controlled evaluation. Start with the repository documentation and test it with authorised sources.

Review the repository