annotation-overlay-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| ANNO_PORT | No | HTTP server port | 3847 |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| read_annotationsA | Read all pending visual annotation feedback. Each annotation includes a CSS selector, fallback selector chain (id → cssPath → contentHash), comment, annotation type (arrow/box/text/freehand/select/textsel), color, viewport position, and element metadata. Use this to consume user feedback after they press Submit in the overlay. |
| clear_annotationsA | Clear all stored annotations. Call after processing feedback to prepare for the next annotation round. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 2 tools
read_annotations and clear_annotations have completely distinct purposes: one retrieves feedback, the other removes it. There is no overlap or ambiguity between the two operations.
Both tools follow the same verb_noun pattern (read_annotations, clear_annotations), with clear and predictable naming. The convention is consistent across the entire server.
With only 2 tools, the server feels minimal. While the read/clear pair covers the core workflow, the count is borderline for a standalone MCP server.
The server covers the essential operations for consuming and resetting annotation feedback. There is no per-annotation update or delete, but the batch read/clear model fits the described usage pattern well.