[restartcosta]

Add Stilingue Social Listening MCP server

This PR adds a single new partner entry, partners/servers/stilingue-mcp-server.json, registering the Stilingue Social Listening MCP server for the Microsoft Foundry Tools Catalog and the Azure API Center MCP Registry. No other files are touched.

---

🦉 What Stilingue is

Stilingue is a social listening and customer care platform from Blip, used by advertising, marketing and customer service teams to collect social media publications through the networks' APIs and turn them into analysis for their clients.

This MCP server puts that platform inside an AI agent. Instead of exporting data from a panel, pasting it into a spreadsheet and building a deck by hand, an agent can ask Stilingue directly and get real, current numbers back.

"What are people saying about our brand today?" · "Compare us with a competitor over the same period." · "Is my panel well configured?"

What agents can do with it:

	Capability	Example
🔎	Public social listening (Radar)	Boolean search over a public base: posts, volume over time, breakdown by channel, gender, device and sentiment, top influencers
📊	Panel analytics (Warroom)	Publications, trends, sentiment and interactions from a customer's dedicated panel, filtered by channel, connected page and date range
🧭	Panel discovery	List the accounts, panels and connected social pages the signed-in user can access, and flag pages whose connection expired
🛠️	Setup audit and improvement	Review groups, themes, tags and automation rules, then create or edit collection descriptors with a mandatory preview
🎧	Customer service reporting	Smartcare cases, response and handling times, SLA, backlog and operators

How to try it

1. Connect an MCP client to https://mcp.stilingue.com.br/mcp. The OAuth flow opens the Stilingue login page; sign in with the account above.
2. Call list_panels to see the accounts and panels the test user can access.
3. Call radar for a public query, for example a brand name with a date_range such as 202609010000:202609152359.
4. Call warroom_query or setup with the panel apikey above, or with the account_id and universe_id returned by list_panels.

---

✅ Validation

Validated locally against partners/server-schema.json:

$ python -m jsonschema partners/server-schema.json -i partners/servers/stilingue-mcp-server.json
(no errors)

• Every required property from partners/server-schema.json is present.
• maxLength, pattern and enum constraints were checked, including summary (single line, under 200 characters) and description (under 1000 characters).
• categories uses the supported value Analytics.
• vendor is Partner.

Checklist for the repository workflow (.github/workflows/validate-servers.yml):

• check-jsonschema against the full partners/servers/ directory
• Server name uniqueness across JSON files
• Security scheme key uniqueness across JSON files (stilingueoauth2)
• Every URL in the entry (icon, license, documentation, remote, OAuth endpoints, support) returns HTTP 200

---

📬 Contact

Stilingue Support, suporte.sti@blip.ai. Happy to make any changes the review process calls for.

# Add Stilingue Social Listening MCP server

This PR adds a single new partner entry, partners/servers/stilingue-mcp-server.json, registering the Stilingue Social Listening MCP server for the Microsoft Foundry Tools Catalog and the Azure API Center MCP Registry. No other files are touched.

---

🦉 What Stilingue is

Stilingue is a social listening and customer care platform from Blip, used by advertising, marketing and customer service teams to collect social media publications through the networks' APIs and turn them into analysis for their clients.

This MCP server puts that platform inside an AI agent. Instead of exporting data from a panel, pasting it into a spreadsheet and building a deck by hand, an agent can ask Stilingue directly and get real, current numbers back.

"What are people saying about our brand today?" · "Compare us with a competitor over the same period." · "Is my panel well configured?"

What agents can do with it:

	Capability	Example
🔎	Public social listening (Radar)	Boolean search over a public base: posts, volume over time, breakdown by channel, gender, device and sentiment, top influencers
📊	Panel analytics (Warroom)	Publications, trends, sentiment and interactions from a customer's dedicated panel, filtered by channel, connected page and date range
🧭	Panel discovery	List the accounts, panels and connected social pages the signed-in user can access, and flag pages whose connection expired
🛠️	Setup audit and improvement	Review groups, themes, tags and automation rules, then create or edit collection descriptors with a mandatory preview
🎧	Customer service reporting	Smartcare cases, response and handling times, SLA, backlog and operators

Why this is a useful addition to the catalog: it brings a category that is not covered today, live social media listening data with sentiment and influencer analysis, into Foundry agents. Data is returned as structured results scoped to the signed-in user, so the agent answers from measured volumes instead of guessing from web excerpts.

• Hosted endpoint: https://mcp.stilingue.com.br/mcp (streamable HTTP, no install required)
• Docs: https://mcp.stilingue.com.br/home
• Support: suporte.sti@blip.ai

---

🧰 Tools exposed

The server exposes 10 tools. Every tool declares MCP annotations, and only the two that write are marked non-read-only.

Tool	What it does	Access
radar	Search the public Radar base: posts, linechart, statistics, trend_users	Read-only, open world
radar_dashboard	Optional interactive dashboard for the same query (MCP Apps). Text output in hosts without UI support	Read-only, open world
warroom_query	Query a dedicated panel (publications, trends, sentiment, interactions and more)	Read-only
list_panels	Accounts, panels and permissions of the signed-in user	Read-only
list_connected_pages	Connected proprietary social pages, with expired connections flagged	Read-only
setup	Read a panel's configuration: groups, themes, tags, automation rules and collection limits	Read-only
edit_group_descriptors	Add, remove or replace descriptors, rename, or create a collection group	Write, dry-run first
edit_theme_descriptors	Same, for classification themes	Write, dry-run first
smartcare_report	Customer service case report: volume, times, SLA and status history	Read-only
list_operators	Service operators of a panel	Read-only

Important

Writes are never applied directly. edit_group_descriptors and edit_theme_descriptors default to dry_run=true. The server instructions require the agent to show the preview to the user and call again with dry_run=false only after explicit confirmation.

Note

Tool descriptions and server instructions are written in Brazilian Portuguese, and the platform's data comes mainly from Portuguese-language social media.

---

🔐 Authentication

The entry declares one scheme, live in production today:

Scheme	Detail
stilingueoauth2	OAuth 2.1 authorization code + PKCE, with Dynamic Client Registration. Endpoints published at /.well-known/oauth-authorization-server and /.well-known/oauth-protected-resource

• Users sign in with their Stilingue account on a login page hosted by Stilingue, so the password is never handed to the agent or the MCP client.
• The server defines no OAuth scopes. Access is determined by the signed-in user's own permissions on the Stilingue platform. warroom_query and setup verify that the session can access the requested panel.
• Access tokens are refreshable through the refresh_token grant, and /oauth/revoke is available for disconnecting.
• A Stilingue account is required to call the tools. Test credentials for reviewers are listed in [Test access](#-test-access-for-reviewers) below.

---

📦 The entry at a glance

Key fields of stilingue-mcp-server.json

Field	Value
name	stilingue-social-listening-mcp
title	Stilingue Social Listening
kind	mcp
remote / remoteType	https://mcp.stilingue.com.br/mcp / streamable-http
vendor	Partner
categories	Analytics
authSchemas	OAuth2
securitySchemes key	stilingueoauth2
versionName	original
customProperties.x-ms-preview	false
useCases	8 entries

The entry is marked as generally available (x-ms-preview: false). All ten tools, including smartcare_report and list_operators, have been tested and are live in production.

---

🧪 Test access for reviewers

Everything except the login page requires a Stilingue account, so a test account is provided for reviewing this entry.

Login (OAuth)	claude.test@stilingue.com.br
Password	Claude@Sti2026
Panel apikey (Warroom, Setup)	6370889063006208

How to try it

1. Connect an MCP client to https://mcp.stilingue.com.br/mcp. The OAuth flow opens the Stilingue login page; sign in with the account above.
2. Call list_panels to see the accounts and panels the test user can access.
3. Call radar for a public query, for example a brand name with a date_range such as 202609010000:202609152359.
4. Call warroom_query or setup with the panel apikey above, or with the account_id and universe_id returned by list_panels.

---

✅ Validation

Validated locally against partners/server-schema.json:

$ python -m jsonschema partners/server-schema.json -i partners/servers/stilingue-mcp-server.json
(no errors)

• Every required property from partners/server-schema.json is present.
• maxLength, pattern and enum constraints were checked, including summary (single line, under 200 characters) and description (under 1000 characters).
• categories uses the supported value Analytics.
• vendor is Partner.

Checklist for the repository workflow (.github/workflows/validate-servers.yml):

• check-jsonschema against the full partners/servers/ directory
• Server name uniqueness across JSON files
• Security scheme key uniqueness across JSON files (stilingueoauth2)
• Every URL in the entry (icon, license, documentation, remote, OAuth endpoints, support) returns HTTP 200

---

📬 Contact

Stilingue Support, [suporte.sti@blip.ai](mailto:suporte.sti@blip.ai). Happy to make any changes the review process calls for.