A comprehensive MCP server providing AI assistants with full access to Cisco Webex messaging capabilities including messages, rooms, teams, people, webhooks, and enterprise features.
Static source inference · medium confidence · detected: Logging
Deprecated protocol patterns detected
Summary
This server exhibits critical, systemic definition quality failures across nearly all 52 tools. The most severe issue is that 41 out of 52 tools (79%) have EMPTY input schemas, just `{"type":"object","properties":{},"required":[]}` with no parameters defined. For example, 'create_attachment_action' has description 'Create a new attachment action in Webex.' with zero parameters documented, an agent cannot invoke this tool. The server provides tool names and basic descriptions, but lacks the detailed parameter documentation, output schemas, error handling guidance, and LLM-optimized descriptions required by production standards. Naming is acceptable (verb_noun convention), but cannot compensate for missing schemas and descriptions.
CRITICAL: Add complete parameter schemas to all 41 tools with empty schemas. For each tool, document: parameter name, type (string, integer, boolean, object, array), description (50-100 chars explaining what it controls), constraints (enum, min/max, regex pattern, format), required boolean, and default value where applicable. Example: 'roomId' (string, required, 'The unique Webex room identifier, use list_rooms to discover available IDs'), 'limit' (integer, optional, min=1, max=100, default=20, 'Number of results to return per page').
CRITICAL: Expand all tool descriptions from 15-20 chars to 100-200 chars following LLM-optimized guidelines. Use: WHAT (one sentence action), WHEN (when to use vs similar tools), PREREQUISITES (lookup first? which tools?), OUTPUT (what data returns). Example for create_message: 'Send a text or rich message to a Webex room. Use this to post responses, confirmations, or summaries. Requires a valid roomId, use list_rooms to find available rooms. Returns a message object with messageId for reference in edits or replies.'
CRITICAL: Document output schemas for all 52 tools. For each tool, list the returned fields, types, and what they represent. Example for get_room_details: 'Returns {roomId: string, roomName: string, description: string, isTeamRoom: boolean, createdAt: ISO8601, memberCount: integer}', enable agents to know what data is available for downstream tool calls.
CRITICAL: Add parameter descriptions to all tools (currently only create_membership has partial parameter documentation). Use the schema 'description' field to explain constraints and usage. Example: 'personEmail (string, required if personId omitted): Email address of the person to add. Must be a valid Webex user email in format [email protected]. Cannot be used with personId simultaneously.'
Spec posture evidence
Inferred effective spec: <=2025-11-25.
Relies on Logging (deprecated) - log to stderr or use OpenTelemetry
Score history
Overall score trend
↓ 2 points across a rubric change (v1 → v2)
30/100
Scored
Grade
Overall
Spec posture
Rubric
2026-09-21
F
30
<=2025-11-25
v2
2026-03-09
F
32
-
v1
create_team_membershipwriteauth20/100
Create a team membership in Webex.
create_webhookwriteauth20/100
Create a webhook in Webex.
delete_membershipdestructiveauth20/100
Delete a membership by ID in Webex.
delete_messagedestructiveauth20/100
Delete a message in Webex by message ID.
delete_persondestructiveauth20/100
Delete a person from the system.
delete_roomdestructiveauth20/100
Delete a room in Webex by its ID.
delete_room_tabdestructiveauth20/100
Delete a Room Tab in Webex.
delete_teamdestructiveauth20/100
Delete a team by ID in Webex.
delete_team_membershipdestructiveauth20/100
Delete a team membership by ID in Webex.
delete_webhookdestructiveauth20/100
Delete a webhook by its ID from the Webex Messaging API.
edit_messagewriteauth20/100
Edit a message in Webex.
get_attachment_action_detailsread onlyauth20/100
Get details for an attachment action by ID.
get_ecm_folder_detailsread onlyauth20/100
Get details for a room ECM folder with the specified folder ID.
get_event_detailsread onlyauth20/100
Get details for a specific event by event ID.
get_membership_detailsread onlyauth20/100
Get details for a membership by ID.
get_message_detailsread onlyauth20/100
Get details of a message by its ID.
get_my_own_detailsread onlyauth20/100
Get profile details for the authenticated user.
get_person_detailsread onlyauth20/100
Get details of a person by their ID from Webex Messaging API.
get_room_detailsread onlyauth20/100
Get details of a room by ID.
get_room_meeting_detailsread onlyauth20/100
Get details of a Webex meeting for a specific room.
get_room_tab_detailsread onlyauth20/100
Get details for a Room Tab in Webex.
get_team_detailsread onlyauth20/100
Get details for a team by ID.
get_team_membership_detailsread onlyauth20/100
Get details for a team membership by ID.
get_webhook_detailsread onlyauth20/100
Get details of a webhook by its ID.
list_direct_messagesread onlyauth20/100
List all messages in a 1:1 room.
list_ecm_folderread onlyauth20/100
List the ECM folder of a specified room in Webex.
list_eventsread onlyauth20/100
List events in your organization using the Webex Messaging API.
list_membershipsread onlyauth20/100
List memberships in a Webex room.
list_messagesread onlyauth20/100
List messages in a Webex room.
list_peopleread onlyauth20/100
List people in your organization using the Webex Messaging API.
Nearly all tool descriptions are trivial: 15-20 characters (e.g., 'Create a new attachment action in Webex.' or 'Get details for a membership by ID.'). These do not explain WHAT the tool does, WHEN to use it vs similar tools, or what parameters/output to expect, leaving agents unable to select the right tool or understand required inputs.
No parameter descriptions provided for any of the 11 tools with non-empty schemas (only create_membership, get_message_details, and others have partial parameter documentation via the 'description' key).
No output schemas documented for any tool. LLMs need to know what fields to expect so they can plan downstream tool calls and extract the right data.' Agents cannot chain tools without knowing what fields the previous tool returned.
Tool naming inconsistency: some tools use 'update_a_X' (e.g., 'update_a_membership', 'update_a_person') while others use 'update_X' (pattern:tool names should follow consistent verb_noun convention). This forces LLMs to reason about naming variations when they expect consistent patterns.
No pagination support documented. List tools (list_rooms, list_messages, list_people, etc.) do not specify whether they support pagination, limit, or offset parameters.
No identification of which parameters are mutually exclusive, required, or conditional. For example, if create_membership requires EITHER personId OR personEmail but not both, this is not documented.
create_membershipAll WRITE tools with multiple optional parameters
No distinction between tools that return sensitive data vs. public data, and no guidance on whether credentials/tokens could leak.
All GET and LIST tools
Add pagination support to all list_* tools. Define limit (1-100, default 20), offset (0+, default 0), and nextCursor fields in output. Example for list_messages: 'Returns {messages: [{messageId, text, authorId, createdAt}], total: integer, nextCursor: string|null}, pass nextCursor as the offset parameter to fetch the next page.'
Add confirmation or dry-run support to all destructive tools (delete_*). Use an optional 'dryRun' boolean parameter (default false) that returns what would be deleted without committing the change. Example: 'delete_message(messageId, dryRun=true)' returns '{wouldDelete: {messageId, text, author}, needsConfirmation: true}' so agents can confirm before irreversible deletion.
Document mutually exclusive and conditional parameter relationships in descriptions. Example for create_membership: 'Provide EITHER personId (exact match) OR personEmail (case-insensitive lookup), not both. If both supplied, personId takes precedence. ModeratorStatus is optional and defaults to false.'
Add error handling guidance to all tools. Document 3 categories: RETRYABLE (network timeout, rate limit, retry with backoff), USER_FIXABLE (invalid ID format, missing required field, prompt user to correct), FATAL (permission denied, invalid credentials, stop and explain). Example: 'Possible errors: {code: "NOT_FOUND", message: "Room ID does not exist. Try list_rooms to find valid room IDs.", retryable: false}, {code: "RATE_LIMITED", message: "Too many requests. Wait 60s and retry.", retryable: true}'.
Standardize naming to verb_noun pattern. Rename 'update_a_membership' → 'update_membership', 'unlink_an_ecm_linked_folder' → 'unlink_ecm_folder', 'update_an_ecm_linked_folder' → 'update_ecm_folder'. This removes ambiguity and aligns with baseline conventions.
Document API rate limits and timeouts for each tool. Specify: requests per second, burst limits, and recommended backoff strategy. Example: 'Rate limited to 100 calls/min. On 429 response, exponential backoff with jitter (wait 1s, 2s, 4s, etc.) up to 60s.'
Add examples of valid input and expected output for 5-10 high-impact tools (create_message, create_room, list_messages) in extended documentation, separate from the description field. This helps agents understand realistic usage without polluting the parameter schema.
For tools that can return large result sets (list_messages, list_people), explicitly cap results in the description and explain pagination. Example: 'Returns up to 50 results per call (configurable via limit parameter). Larger result sets must be fetched using pagination, pass the nextCursor from one response as the offset parameter for the next call.'
Add tool annotations (readOnlyHint, destructiveHint, idempotentHint from MCP spec 2026-07-28) to tool definitions. Mark read-only tools (get_*, list_*) with readOnlyHint=true. Mark destructive tools (delete_*) with destructiveHint=true. Mark idempotent tools with idempotentHint=true so clients can optimize retries and safety.
For complex multi-step workflows (e.g., create a team, add members, set permissions), document the recommended tool sequence in extended docs. Example: 'Typical workflow: (1) create_team → get teamId, (2) create_team_membership → add members, (3) update_team_membership → set moderators'. This guides agents in planning multi-round tasks.