MCP server that stores and recalls your AI context as Markdown files in storage you own — local disk, your Dropbox, or your Google Drive.
Contextly has 5 tools with clear verb-noun naming (save_, append_, list_, recall_, search_). All tools have descriptions (avg 120 chars, within baseline 34-392). Input schemas are present with typed parameters. However, parameter descriptions are sparse or missing context about constraints, formats, and error recovery. Output schemas are not documented. Error handling lacks guidance on what to do when a context is not found or when storage fails. The tool set is well-composed (each does one thing), but lacks the depth of documentation expected for production use.
Append new content to an existing context (creates it if missing).
List everything stored in Contextly memory as a table with details: created date, last updated, the model/app it came from, the device, and size. Render the table as-is.
Retrieve a stored context by name so you can use it in the current conversation.
Save (or overwrite) a named context so it can be recalled later from any model. Pass the `model` you are if you know it (e.g. 'Claude Opus 4.7') so the library shows where each context came from.
Search across all stored contexts for a keyword or phrase; returns snippets.
Output schemas not documented. LLMs cannot plan downstream operations or extract structured data from list_contexts, recall_context, or search_context responses.
Parameter descriptions lack format/constraint details. 'title' and 'content' parameters have minimal guidance on length limits, character restrictions, or Markdown dialect expectations.
Error handling lacks recovery guidance. No documentation of what happens when a context is not found, storage quota is exceeded, or network access to Dropbox/Google Drive fails.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 55 | 2026-07-28+ | v2 |
list_contexts returns 'a table with details' but output structure is not formally specified. LLMs cannot reliably parse or extract fields from the response.
search_context returns 'snippets' but snippet format, length, and structure are undocumented. No pagination or result limit guidance.