Return the full sitemap of Adobe VIP Marketplace API documentation. Use this to browse all available topics or to find exact page paths.

Search the Adobe VIP Marketplace API documentation by keyword or topic. Ranks pages by title/tag matches plus content matches (for any page already in the local cache — call warm_vipmp_cache once to pre-fetch every page for the strongest results).

Returns matching page titles, paths, and the most query-relevant sections from each page (not a blunt character cutoff).

Args: query: Search terms, e.g. "create order", "oauth token", "3YC eligibility" max_results: Maximum number of pages to return (default 5, max 10)

Fetch the full content of a specific Adobe VIP Marketplace documentation page.

Args: path: The doc path, e.g. "/vipmp/docs/order_management/create_order/" Use list_vipmp_docs() or search_vipmp_docs() to find valid paths.

Pre-fetch every page in the sitemap so subsequent searches can do content-level matching (not just title/tag matching).

First run takes ~30 seconds over ~70 pages. Subsequent runs are near-instant because the disk cache + ETag revalidation handle unchanged pages with 304 responses.

Returns a summary of fetches, cache hits, and any errors.

Rebuild the sitemap from Adobe's published /sitemap.xml. Fetches every current /vipmp/docs/ page for its title, merges hand-curated search tags where paths still line up, and persists the result to sitemap.json so future server startups use the refreshed list.

Run this if:

• You're seeing 404s on known pages (Adobe renamed or removed them)

• You want to pick up newly-added pages Adobe's published since install

• The default hand-curated sitemap looks stale

Takes about 30 seconds over ~80 pages. Safe to run repeatedly.

Return statistics about the on-disk docs cache — total entries, fresh/stale split, TTL, and the cache file path. Useful when debugging "why did it return stale content" or "is the cache being used at all".

Clear the docs cache. Pass a specific doc path to invalidate one entry, or omit to wipe the entire cache.

Args: path: Optional docs path to invalidate (e.g. "/vipmp/docs/lga/create/"). If None, clears everything.

Extract every HTTP endpoint (method + path) documented across the whole VIPMP docs. Returns a Markdown table grouped by resource area.

Served from the pre-built index (milliseconds) when available. Falls back to live extraction across the sitemap (~30s cold, ~5s warm) if no index is present. Call rebuild_vipmp_index to refresh locally.

Extract every error code documented across VIPMP docs — both numeric codes (e.g. 1117, 5117) and symbolic reason codes (e.g. INVALID_LM_MIGRATION_LEVEL).

Args: query: Optional substring filter. Matches against code, reason, or endpoint. Case-insensitive.

Served from the pre-built index when available, falling back to live extraction if not. Call rebuild_vipmp_index to refresh locally.

Extract every resource status code documented across VIPMP docs (numeric codes 1000-1026). These describe the lifecycle state of a resource — e.g. account, order, or subscription status — as opposed to the request-failure error codes surfaced by list_vipmp_error_codes.

Args: query: Optional substring filter. Matches against code, description, or applicable resources. Case-insensitive.

Served from the pre-built index when available, falling back to live extraction if not. Call rebuild_vipmp_index to refresh locally.

Return structured field schemas (name, type, required, description) for VIPMP resources. Source: the Resources & Fields reference page.

Args: resource_name: Optional case-insensitive substring to filter by resource name (e.g. "customer", "linkedMembership"). Omit to get every documented resource.

Extract code examples (JSON / curl / bash / python / etc.) from a specific docs page. Optionally filter by language.

Args: docs_path: Docs path (e.g. "/vipmp/docs/customer-account/create-customer-account"). language: Optional filter — "json", "python", "bash", etc. Matches the language-<x> class on <code> elements.

Rebuild the pre-extracted index of endpoints, error codes, status codes, and schemas by walking every page in the current sitemap. Saves to ~/.cache/swo-adobe-vipm-docs-mcp/index.json.

Run this if:

• The structured tools (list_vipmp_endpoints, list_vipmp_error_codes, get_vipmp_schema) are returning stale data

• You just ran refresh_vipmp_sitemap and want the structured tools to pick up new pages

• The package-shipped baseline index is older than you'd like

Takes ~60 seconds from cold cache, ~15-20 seconds from warm cache.

Return structured VIPMP release notes — the highest-signal information for developers tracking API changes. Served from the pre-built index (refreshed daily), so calls return in milliseconds.

Each entry has a date, a section ("api_changes", "sandbox", "upcoming", or "earlier"), and one or more changes with titles and bodies. Covers both the main API changes and the Sandbox-specific changes at the bottom of the release-notes page — they're tracked separately.

Args: since: ISO date ("YYYY-MM-DD"). If provided, returns only releases on or after this date. Example: "2026-01-01" for everything since New Year. Entries without a date (catch-all "earlier releases" buckets) are excluded when since is set. section: Optional filter. One of: - "api_changes" — the main API release stream - "sandbox" — Sandbox-environment-specific changes - "upcoming" — announced but not yet shipped - "earlier" — catch-all entries without a specific date limit: Maximum number of release entries to return (default 20).

Examples: # What shipped in 2026? list_vipmp_releases(since="2026-01-01", section="api_changes")

# Recent sandbox-only changes
list_vipmp_releases(section="sandbox", limit=5)

# What's coming
list_vipmp_releases(section="upcoming")

Return SoftwareOne-authored operational tips for topic — rules, gotchas, and field notes that complement Adobe's reference docs.

Topic matching is case-insensitive and punctuation-tolerant: "customer lifecycle", "Customer Lifecycle", and "customer_lifecycle" all resolve to the same section. Use list_vipmp_tip_topics to see what topics exist.

Args: topic: The topic to fetch tips for (e.g. "customer lifecycle", "ordering flow", "3YC", "auth and sandbox").

Returns the raw tips markdown verbatim. Empty-topic and missing-topic cases return a short placeholder message rather than an error, so the assistant can explain gracefully when a topic has no tips yet.

Return the list of topics for which SoftwareOne tips are available.

Each entry maps to an H2 section in the tips file; call get_vipmp_tips(topic) to read one. Useful as a discovery call — "what tips do you have?" — before narrowing to a specific topic.

Dump diagnostic info about the running server — useful as the first call when debugging "why is this not working" or "what version am I actually on".

Returns package version, Python version, index age + counts, sitemap size, cache stats, and log file path.

One-shot profile of a VIPMP endpoint. Returns schema, error codes, code examples, and any release-note mentions in one call — so Claude doesn't have to chain 4 separate tool invocations.

Args: method: HTTP method (GET / POST / PATCH / PUT / DELETE). path: API path (e.g. "/v3/customers").

Example: describe_vipmp_endpoint(method="POST", path="/v3/customers")

Cross-check a JSON request body against the documented VIPMP schema.

Finds the endpoint's schema in the pre-built index and checks every field against its documented type, required-ness, and constraints (character limits, numeric ranges). Flags unknown fields, missing required fields, type mismatches, constraint violations, and deprecated-field usage.

Scope: top-level fields only. Nested objects are noted as "not recursively validated" — check their schemas separately via get_vipmp_schema.

Args: endpoint: "METHOD /path" (e.g. "POST /v3/customers"). body_json: The request body to check, as a JSON string.

Example: validate_vipmp_request( endpoint="POST /v3/customers", body_json='{"resellerId": "5556667778", "externalReferenceId": "342"}', )

Emit a runnable code snippet for a VIPMP endpoint.

Useful for "give me a starting point for calling POST /v3/customers". If you supply a body, it's used as-is. If not, we build a placeholder body from the schema, filling required fields with type-appropriate dummy values so the snippet shows the minimum valid shape.

Args: endpoint: "METHOD /path" (e.g. "POST /v3/customers"). body_json: Optional JSON body to embed in the snippet. If omitted and the method is POST/PATCH/PUT, a placeholder body is constructed from the schema. language: Output language. One of: "curl" (default), "powershell", "python" (httpx), "csharp" (HttpClient).

Example: generate_vipmp_request( endpoint="POST /v3/customers", language="python", )