MCP server for Bugzilla — search, discuss, and manage Firefox/Mozilla bugs via the Bugzilla REST API.
Server has 17 tools with consistent naming (verb_noun pattern) and basic descriptions. All tools have input schemas with type definitions and parameter descriptions. However, descriptions are often terse (10-50 chars), output schemas are not documented, and error handling lacks recovery guidance. No tool annotations (readOnlyHint/destructiveHint). Parameter descriptions are minimal, many lack format/constraint details. No pagination guidance in search results. Follows ~40% of the 54 patterns.
Create an attachment on a bug
Create a new bug in Bugzilla
Add a comment to a bug
Fetch attachments for a bug
Fetch a single bug by ID or alias
Fetch available bug fields
Fetch the history of changes for a bug
Output schemas not documented. LLMs cannot plan downstream calls or extract required fields (e.g., bug_id for chaining). Responses are JSON-stringified but structure is opaque to the agent.
Tool annotations missing. No readOnlyHint, destructiveHint, or idempotentHint. Agents cannot distinguish safe reads from destructive writes (create_bug, update_bug, delete operations).
Error handling lacks recovery guidance. Errors return generic messages ('failed to create bug') without actionable next steps. No distinction between retryable, user-fixable, and fatal errors.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 58 | 2026-07-28+ | v2 |
Fetch comments on a bug
Fetch flag types for a product/component
Fetch details for a specific product
Fetch list of products
Fetch user information by name/email
Search Bugzilla bugs with filters. Use quicksearch for full-text queries.
Search for comment tags
Update an attachment
Update an existing bug
Update tags on a comment
Parameter descriptions are terse (10-50 chars). Many lack format constraints, ranges, or examples. E.g., 'severity' has no enum values; 'priority' has no valid range (P1-P5 inferred but not stated).
search_bugs lacks pagination guidance. No mention of result limits, offset behavior, or total count in response. Large result sets could exhaust context window.