sf-architect-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| SF_ARCHITECT_DB_DIR | No | Override the database directory path. Defaults to ~/.sf-architect-mcp. | ~/.sf-architect-mcp |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {} |
| prompts | {} |
| resources | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| scrape_fullA | Full refresh: wipe all indexed data, re-fetch the sitemap from architect.salesforce.com, and re-scrape and reindex every page. Use this for initial setup or when you want a clean slate. |
| scrape_incrementalA | Incremental scrape: fetch the sitemap, scrape any new pages not previously indexed, retry previously failed pages, and re-check existing pages for content changes. Only re-indexes pages whose content has actually changed. |
| search_architect_docsA | Search the locally indexed Salesforce Architect documentation. Returns ranked results with excerpts. The database must be populated first via scrape_full or scrape_incremental. |
| read_architect_pageA | Read the markdown content of a specific indexed Salesforce Architect page by its URL. Use max_chars to limit output size. For a lightweight overview without full content, prefer read_architect_page_summary instead. |
| read_architect_page_summaryA | Get a lightweight summary of an indexed page: title, section, word count, section headings, and a short preview (~500 chars). Use this to understand a page before deciding to read the full content. |
| get_section_summaryA | Get a quick overview of a section: page count, total word count, and list of page titles. Useful for orientation before exporting or reading individual pages. |
| list_architect_sectionsB | List all indexed sections with page counts and titles. Optionally filter to a specific section. |
| export_architect_sectionA | Export all pages in a section as a single concatenated markdown file. Writes to disk with zero context window cost. When output_path is omitted, defaults to ~/.sf-architect-mcp/exports/{section}-{language}.md. |
| get_scrape_statusA | Get the current status of the indexed database: total pages, chunks, sections breakdown, last scrape run info, and pending/failed URL counts. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| scrape-docs | Scrape Salesforce Architect documentation into the local database. Guides you through language selection before starting. |
| research-topic | Research a Salesforce architecture topic using the locally indexed documentation. Searches, summarizes, and synthesizes findings with source citations. |
| export-section | Export a documentation section to a single markdown file on disk. |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| usage-guide | Tool usage guide — recommended workflows, when to use each tool, and tips for working with the SF Architect documentation. |
| available-languages | Language codes accepted by scrape, list, and export tools, with display names and URL segments. |
| indexed-sections | Currently indexed documentation sections with page counts (live from database). |
TDQS
Scored across 9 tools
Most tools target distinct operations (scrape, search, read, export, status). The only real overlap is between get_section_summary and list_architect_sections, which both return section page counts and titles, and read_architect_page vs read_architect_page_summary, though the latter pair is clearly differentiated by the descriptions.
All nine tools follow a consistent snake_case verb_noun pattern (scrape_full, search_architect_docs, read_architect_page, export_architect_section, get_scrape_status). Resources and prefixes (architect, section, scrape) are used predictably.
Nine tools is well-scoped for a documentation indexing/retrieval server, covering ingestion (2 scrape modes), discovery (sections), retrieval (search, read, read summary), export, and status with no redundancy.
The surface covers the full lifecycle: full/incremental ingestion, status monitoring, section listing and summaries, search, page reading, and export. No obvious dead ends, and scrape_full handles the reset case that a delete tool would otherwise cover.