Contents
What is it?
The Filtered Copilot MCP connector brings your Filtered catalogue to Microsoft 365 Copilot. You can search your Filtered organisation's assets, pathways, skills and internal content directly from Copilot.
How can I install it?
In Copilot chat
Your organisation admin can enable or install the Filtered connector for you. If it’s not installed but enabled in your organisation you can Add it for yourself inside https://m365.cloud.microsoft/chat
Using the connector requires Filtered access for your company email address. Even if your Microsoft 365 admin has already added the connector, each user still needs to connect it themselves: follow the “Connect the connector with OAuth” section of this documentation, and Copilot will guide you through the OAuth sign-in.
Add the connector to as a “Data source”

If you or your admin have already added it, you'll find it at https://m365.cloud.microsoft/chat → Settings → Sources → Filtered.

Connect the connector with OAuth
In the Filtered connector, click "Connect" to link your Filtered account to Microsoft 365 Copilot.

To connect, you need to sign in to Filtered. Click "Continue to Filtered" and you'll be redirected to the Filtered sign-in screen. Connecting lets Filtered provide personalised results inside Microsoft 365 (Copilot, Word, Excel and more) and Teams.

On the Filtered sign-in screen, first enter the email address registered with Filtered. On the next screen, choose your usual Filtered sign-in method.

Once you've signed in, you'll be redirected to the Copilot settings screen with a "Filtered is now connected" message.

After connecting, you should see the Filtered MCP tools:

Try the Filtered connector in Copilot Chat
Ask Copilot to query your Filtered data, such as internal content, pathways, skills, ingested SCORM packages and job roles.

(For admins) In MS365 admin center
Add the connector to a Microsoft 365 tenant
This requires Microsoft 365 administrator privileges. If you can't see the connector in Copilot Chat → Settings → Sources, ask your admin to enable it for you.
- Select the Filtered connector in MAC → Copilot → Connectors

- Click "Sign in" to check that the connection works. This starts the OAuth consent flow and requires Filtered platform access.

- Complete the OAuth consent flow

- Make sure to enable the connector’s results to be displayed in Copilot

Filtered Connective Intelligence MCP Tools
This MCP server gives AI assistants read-only access to a tenant's Filtered Connective Intelligence catalogue: learning assets, skills, job roles, pathways and quality data.
Every tool is read-only (readOnlyHint: true) and scoped to the calling tenant, so you only ever see your own catalogue.d
Tool results are the Filtered API JSON body, returned as-is. The examples below show the parsed tool payload, not the outer MCP JSON-RPC envelope.
Quick reference
Tool | What it does | Required params |
search | Search the catalogue with filters | q |
get_assets | List assets, paginated | none |
get_asset | Get one asset by id | asset_id |
get_asset_quality | Quality scores for one asset | asset_id |
get_asset_languages | Languages present in the catalogue | none |
get_asset_formats | Formats present in the catalogue | none |
get_asset_providers | Providers present in the catalogue | none |
get_skills | List all skills | none |
get_skill_groups | Skill groups with their skills | none |
get_skill_labels | Skill labels with their skills | none |
get_job_roles | Job roles with their skills | none |
get_tagged_assets | Assets tagged with a skill | skill_id |
get_tagged_assets_by_label | Assets tagged via a skill label | label_id |
get_tagged_assets_by_job_role | Assets tagged via a job role | job_role_id |
get_pathways | List all pathways | none |
get_pathway | Get one pathway with sections and assets | pathway_id |
Authentication
The server uses OAuth. See Errors below for what happens when a token is rejected.
Common workflows
Find content on a topic. Call search with a query. If you want to filter, first check what values actually exist using get_asset_languages, get_asset_formats and get_asset_providers. Filtering on a value that isn't in those lists returns nothing.
Find content for a role. Call get_job_roles to find the role's id, then get_tagged_assets_by_job_role.
Find content for a skill. Call get_skills (or get_skill_groups / get_skill_labels to browse) to find an id, then get_tagged_assets or get_tagged_assets_by_label.
Assess asset quality. Call get_asset for the overall score and labels, then get_asset_quality for the per-metric breakdown.
Explore curated learning paths. Call get_pathways, then get_pathway for the full structure.
Assets
search
Search the catalogue. Returns matching assets, each with a relevance score.
Param | Type | Default | Notes |
q | string | required | The search query |
limit | integer | 15 | Max results |
relevance_threshold | number | 0.0 | Minimum relevance score to include |
use_endorsement | boolean | false | Factor endorsement into ranking |
language | string | null | null | See get_asset_languages |
format | string | null | null | See get_asset_formats |
provider | string | null | null | See get_asset_providers |
endorsed | boolean | null | null | Only endorsed ( true) or non-endorsed (false) assets |
use_endorsementinfluences ranking,endorsedfilters results.
Example request:
{
"q": "stakeholder management",
"limit": 10,
"language": "en",
"endorsed": true
}Example response:
{
"assets": [
{
"id": "72cfb7d7-be0d-4f1b-8b93-65e2725927f1",
"title": "Test title",
"description": "Test description",
"subtitle": null,
"image_url": null,
"type": null,
"length": null,
"url": null,
"provider": null,
"publication_year": null,
"organisation_asset_id": null,
"language": null,
"author": null,
"availability": null,
"is_endorsed": false,
"created_at": null,
"updated_at": null,
"expiration_date": null,
"source": null,
"modified_at_external": null,
"relevance": 0.2,
"overall_quality_score": null,
"labels": [],
"tagged_skills": [],
"tagged_proficiency": null,
"scorm_asset_id": null,
"scorm": null
}
]
}get_assets
Paginated list of all assets. The response includes total for paging.
Param | Type | Default | Notes |
offset | integer | 0 | Start position |
limit | integer | 100 | Page size |
language | string | null | null | Two-letter ISO 639-1 code, e.g. en, de |
Example request:
{ "offset": 100, "limit": 50, "language": "en" }Example response:
{
"total": 150,
"assets": [
{
"id": "420ef543-7796-4af5-a373-ad631f1eeff3",
"title": "Asset title 1",
"description": "Asset description 1",
"subtitle": "Short subtitle for asset 1",
"image_url": "https://example.com/image1.png",
"format": "video",
"length": 120,
"url": "https://example.com/asset1",
"provider": "Provider A",
"publication_year": 2020,
"organisation_asset_id": "external_asset_id_1",
"language": "en",
"author": "Jane Doe",
"availability": "Available",
"is_endorsed": true,
"created_at": "2024-01-15T10:30:00",
"updated_at": "2024-03-20T08:45:00",
"expiration_date": "2027-01-01T00:00:00",
"source": "DEGREED",
"modified_at_external": "2024-03-19T12:00:00Z",
"overall_quality_score": null,
"labels": [],
"tagged_skills": []
}
]
}get_asset
Get a single asset. Returns the same fields as the list, including the overall quality score and labels.
Param | Type | Notes |
asset_id | string | required |
Example response:
{
"id": "72cfb7d7-be0d-4f1b-8b93-65e2725927f1",
"title": "Asset title 1",
"description": "Asset description 1",
"subtitle": "Short subtitle for asset 1",
"image_url": "https://example.com/image1.png",
"format": "video",
"length": 120,
"url": "https://example.com/asset1",
"provider": "Provider A",
"publication_year": 2020,
"organisation_asset_id": "external_asset_id_1",
"language": "en",
"author": "Jane Doe",
"availability": "Available",
"is_endorsed": true,
"created_at": "2024-01-15T10:30:00",
"updated_at": "2024-03-20T08:45:00",
"expiration_date": "2027-01-01T00:00:00",
"source": "DEGREED",
"modified_at_external": "2024-03-19T12:00:00Z",
"overall_quality_score": "3.7",
"labels": ["Beginner", "Compliance"],
"tagged_skills": [
{
"id": "5d3e2f10-8a7b-4c6d-9e1f-2a3b4c5d6e7f",
"name": "Python"
}
]
}overall_quality_scoreis a string (e.g."3.7"), ornullif the asset hasn't been evaluated.
get_asset_quality
Per-metric quality scores and an evaluation summary for one asset.
Param | Type | Notes |
asset_id | string | required |
Unevaluated asset:
{
"quality_metrics": [],
"evaluation_summary": null
}Evaluated asset:
{
"quality_metrics": [
{
"metric": "TOPIC_RELEVANCE",
"name": "Topic relevance",
"order": 1,
"category": "CONTENT",
"score": 5,
"explanation": "Highly relevant",
"improvement_plan": "No changes needed"
},
{
"metric": "CONTENT_FRESHNESS",
"name": "Content freshness",
"order": 2,
"category": "CONTENT",
"score": 3,
"explanation": "Content references Office 2016",
"improvement_plan": "Update to mention Microsoft 365"
}
],
"evaluation_summary": {
"overall_score": "3.7",
"summary": "The package is well structured.",
"priority_recommendation": "Review dated examples.",
"key_strengths": ["Coherent topic coverage"],
"key_weaknesses": ["Outdated references"],
"status": "AMBER",
"status_explanation": "At least one scored metric is 3 or below.",
"evaluated_at": "2026-07-01T12:34:56"
}
}get_asset_languages
No parameters. Returns the distinct languages in your catalogue.
{
"languages": ["de", "en", "en-US"]
}Values aren't always two-letter codes (noteen-US). Use exactly what this returns when filteringsearch.
get_asset_formats
No parameters. Returns the distinct formats in your catalogue.
{
"formats": ["article", "book", "video"]
}get_asset_providers
No parameters. Returns the distinct providers in your catalogue.
{
"providers": ["Provider A", "Provider B"]
}Skills and roles
get_skills
All skills in the catalogue. No parameters.
{
"skills": [
{
"id": "a797da91-e3bf-40e1-a365-16114a38f6d2",
"name": "Tagged asset 1",
"definition": "Tagged asset 1 definition"
},
{
"id": "ff6f5202-7d0e-4ef5-95ff-6ea1724d3348",
"name": "Tagged asset 2",
"definition": ""
}
]
}get_skill_groups
All skill groups, each with its skills. No parameters. The payload is a top-level JSON array, not an object.
[
{
"id": "c80705ae-d943-4d27-afd9-b7075d327f21",
"name": "Example skill group 1",
"skills": []
},
{
"id": "07e2fdec-54ce-4b73-af50-8e7bba972498",
"name": "Example skill group 2",
"skills": [
{ "id": "de06c88c-dc53-441a-b3b2-eaba522cbc5c", "name": "Skill 1" },
{ "id": "a295ac6a-adfc-4fd7-a868-f5340e264e7e", "name": "Skill 2" }
]
}
]get_skill_labels
All skill labels, each with its skills. No parameters.
{
"labels": [
{
"id": "98190f1e-bf86-419a-986f-ef8bd9e39314",
"name": "Data",
"skills": [
{ "id": "4f01b155-14a8-4894-a3c4-fd3a2e66ce37", "name": "SQL" },
{ "id": "6cf9ec6d-e2be-43cd-a14a-e9ad39589e57", "name": "Python" }
]
},
{
"id": "1cfbc608-cd57-4aea-a94e-e1592ab6749a",
"name": "Leadership",
"skills": []
}
]
}get_job_roles
All job roles, each with its skills. No parameters.
{
"job_roles": [
{
"id": "dadf1e95-fded-4e5d-b894-e685fafc032a",
"name": "Data Analyst",
"skills": [
{ "id": "4f01b155-14a8-4894-a3c4-fd3a2e66ce37", "name": "SQL" },
{ "id": "6cf9ec6d-e2be-43cd-a14a-e9ad39589e57", "name": "Python" }
]
},
{
"id": "26cbf095-b7a4-4105-89f2-96ec2e93bbcd",
"name": "Engineering Manager",
"skills": []
}
]
}Tagged assets
These three tools share the same pagination behaviour and response shape: total plus an assets array, where each asset carries a relevance score.
Param | Type | Default |
offset | integer | null | null |
limit | integer | null | null |
If you omit offset and limit you get every tagged asset. For large skills, labels or roles, always pass a limit.
get_tagged_assets
Assets tagged with one skill.
Param | Type | Notes |
skill_id | string | required |
{
"total": 2,
"assets": [
{
"id": "72cfb7d7-be0d-4f1b-8b93-65e2725927f1",
"title": "Asset title 1",
"description": "Asset description 1",
"relevance": 0.95,
"provider": "Provider A",
"organisation_asset_id": "external_asset_id_1",
"is_endorsed": false,
"language": null,
"overall_quality_score": null,
"labels": [],
"tagged_skills": [],
"tagged_proficiency": null
}
]
}get_tagged_assets_by_label
Assets tagged with any skill under a label. relevance is the asset's highest score across the label's skills.
Param | Type | Notes |
label_id | string | required |
{
"total": 1,
"assets": [
{
"id": "72cfb7d7-be0d-4f1b-8b93-65e2725927f1",
"title": "Asset title 1",
"description": "Asset description 1",
"relevance": 0.95,
"overall_quality_score": "4.2",
"provider": "Provider A",
"labels": [],
"tagged_skills": []
}
]
}get_tagged_assets_by_job_role
Assets tagged with any skill of a job role. relevance is the asset's highest score across the role's skills.
Param | Type | Notes |
job_role_id | string | required |
Example request:
{ "job_role_id": "dadf1e95-fded-4e5d-b894-e685fafc032a", "offset": 0, "limit": 25 }Example response:
{
"total": 1,
"assets": [
{
"id": "72cfb7d7-be0d-4f1b-8b93-65e2725927f1",
"title": "Asset title 1",
"description": "Asset description 1",
"relevance": 0.95,
"provider": "Provider A",
"labels": [],
"tagged_skills": []
}
]
}Pathways
get_pathways
All pathways. No parameters. The payload is a top-level JSON array.
[
{
"id": "24cab24c-35c6-11ee-be56-0242ac12024c",
"name": "Test Name 1",
"description": "Some description",
"created_at": "2023-11-02T00:00:00",
"created_by": "Some user",
"number_of_assets": 2,
"full_length": 123
},
{
"id": "ff58834f-d2b2-484e-ab24-4e1b714ae445",
"name": "Test Name 3",
"description": null,
"created_at": "2023-10-01T00:00:00",
"created_by": null,
"number_of_assets": 45,
"full_length": null
}
]get_pathway
One pathway with its sections and assets. Sections come with section_order, and each asset has a position_order within its section.
Param | Type | Notes |
pathway_id | string | required |
{
"id": "83f7d7a1-c03b-418d-9f31-69664f4fea45",
"name": "Mock playlist",
"description": "mock description",
"sections": [
{
"section_id": "39c817fc-35b5-11ee-be56-0242ac12039c",
"title": "Test Section B (order 1)",
"description": "Test Section B description",
"section_order": 1,
"assets": []
},
{
"section_id": "abde17fc-35b5-11ee-be56-0242ac12abde",
"title": "Test Section C (order 3)",
"description": null,
"section_order": 3,
"assets": [
{
"id": "c4815e94-f85e-4cf3-a32c-9a401c9934e6",
"title": "mock-asset-title",
"description": "mock-asset-description",
"subtitle": "mock-subtitle",
"type": "mock-type",
"url": "mock-url",
"provider": "mock provider",
"language": "en",
"length": 120,
"publication_year": 2023,
"organisation_asset_id": "mock-organisation-asset-id",
"position_order": 0,
"rationale": "Rationale",
"required": false,
"is_endorsed": false,
"added_at": "2023-11-23T11:23:50",
"adder_id": "test_user_id",
"adder_name": "Test User",
"created_at": "2023-11-23T11:23:50",
"updated_at": "2024-11-23T11:23:50",
"expiration_date": "2024-03-15T12:34:56",
"author": null,
"availability": null,
"image_url": null,
"source": null,
"modified_at_external": null,
"scorm_asset_id": null
}
]
}
]
}Errors
Authentication failures return 401 Unauthorized with this body:
{
"error": "Unauthorized"
}Token rejected. The response includes this header:
WWW-Authenticate: Bearer error="invalid_token"The access token is invalid or expired. Refresh it and retry. Most integrations (e.g. Copilot) do this automatically.
Missing or malformed header. No WWW-Authenticate header. The request had no Authorization header, or it didn't use the Bearer scheme. Check your integration's configuration; retrying won't help.
Other failures, such as an unknown asset id, come from the Filtered API and surface as tool errors. How they look depends on your MCP client.
