mcp-server

set_ai_source

Store per-product values that YOU (the client's own model) extracted, as a new 'AI source' attribute — so a feed rule can read the attribute directly instead of matching free text with contains/conditions. Typical flow: call preview_products to read products, derive ONE value per product with your own reasoning (e.g. the color, gender, or material from the title/description), then call this with values as a JSON OBJECT (not an array) mapping each product's id (the id field from preview_products) to the extracted value. code is the new attribute name (e.g. "ai_color"); once applied it becomes the source attribute custom_ (visible in list_source_attributes). Send large catalogs in BATCHES: first batch mode:"replace", following batches mode:"append"; set apply:true on the LAST batch only, to materialize (this re-imports the additional sources so the values land on products — an ASYNCHRONOUS import). VERIFY before using it: poll get_import_status until the newest additional_source run in recentRuns has running:false and status 'ok', and the code appears in list_source_attributes; only THEN attach the attribute to a feed with map_feed_attribute. If that run errors, run_import retries without re-sending the values. handle overrides the product key to join on (default header.id, which matches the preview_products id). project_id is OPTIONAL (inferred for a single-project customer). Requires the additional-sources addon: without it the call is refused with error 'addon_required' (HTTP 403) + addonCode + upsellUrl — show the user the upsellUrl so they can unlock it, and do not retry. Returns {status:'written'|'applied', code, url, valueCount, applied}.

Input Schema

TableJSON Schema

Name	Required	Description	Default
`code`	Yes
`mode`	No
`apply`	No
`handle`	No
`values`	Yes
`project_id`	No

TDQS

A4.9/5.0

Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations only say readOnly=false and idempotent=false, so the description carries full burden. It thoroughly discloses: asynchronous re-import on apply, status values ('written'|'applied'), HTTP 403 with addonCode/upsellUrl on missing addon, retry behavior via run_import, and the materialization side effect. No contradiction with annotations.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is long and uses dense run-on sentences, but every clause carries operational necessity. It is front-loaded with purpose and flows logically through batching, verification, and error handling. Slightly more structure would earn a 5, but nothing is wasted.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a complex tool with async imports, batching, and addon requirements, the description covers all critical aspects: precondition flow, verification steps, failure recovery, return shape, and error handling. Even without an output schema, an agent has everything needed to invoke it correctly.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters5/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

With 0% schema description coverage, the description must explain every parameter, and it does: values as JSON object keyed by preview_products id, code becoming custom_<code>, mode replace/append, apply only on last batch, handle overriding the join key, project_id optional. This is far beyond what the bare JSON schema provides.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description opens with a precise verb+resource statement: 'Store per-product values... as a new AI source attribute' and clearly states the purpose (allowing feed rules to read the attribute directly instead of free-text matching). This clearly differentiates it from siblings like delete_ai_source, list_ai_sources, and set_api_source.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Provides a complete workflow: preview_products → derive values → call with replace/append batching → verify via get_import_status and list_source_attributes → then map_feed_attribute. It also gives explicit when-not-to-use (addon_required error means don't retry, show upsellUrl). No ambiguity about when this tool is appropriate.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All mcp-server tools Inspect

Try in Browser

RayrunBuild, deploy, and operate MCP servers without becoming an infrastructure team.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

TDQS

A4.2/5.0

Disambiguation4/5

Tools are organized around distinct resources (ads, marketplaces, feeds, orders, rules, sources) with clear action verbs, and descriptions explicitly disambiguate near-pairs like get_feed_status vs ad_status or set_feed_filter vs set_feed_attribute_filter. A few similarly named status/action pairs (e.g. ad_status vs get_ad, run_ad_item_action vs run_ad_operation) require careful reading, but overall the purposes are separable.

Naming Consistency4/5

The overwhelming majority follow a consistent verb_noun snake_case pattern (list_*, get_*, create_*, set_*, run_*, test_*). Minor deviations like ad_status and marketplace_status (noun-based status tools) and koongo_knowledge break the pattern slightly, but the convention is clearly recognizable and predictable.

Tool Count1/5

At 105 tools, the surface is extreme and far beyond the 50+ threshold, even for a complex e-commerce integration domain. Much of the bloat comes from systematic triplication across ads, marketplaces, and feeds (e.g. three nearly identical map_*_attribute tools, three list_*_items, three get_*_report) that a generic resource parameter could have consolidated.

Completeness4/5

The toolset covers the full lifecycle of feeds, ads, marketplaces, order connections, rules, and imports, including create/read/update/delete, status monitoring, item-level actions, validation, repair, and restore. Minor gaps exist, such as no delete for standalone order connections and limited update capabilities for some entities, but these are workable and do not create dead ends for the core workflows.

Resources

Need Help?