md-log-mcp — Claude Code MCP server: save agent reports
An MCP server that pushes your agent's markdown reports into md-log by path, with versions and images attached.
claude mcp add md-log-mcp -e MDLOG_API_BASE_URL=<MDLOG_API_BASE_URL> -e MDLOG_PAT=<MDLOG_PAT> -- npx -y md-log-mcpWhat md-log-mcp does
Your agent names a path like my-project/2026-07-07-error-report.md and the server sends that file to the hosted md-log service. Folders that don't exist yet get created. Every save lands as a new immutable version, so the edit history stays readable.
There are 17 tools. The headline one is save_markdown: it creates or overwrites a file, uploading embedded images first and swapping placeholders for asset:// links. edit_markdown handles literal replacements inside a file, append_to_markdown adds to the end. Both use optimistic concurrency via base_version_no and retry once on a conflict.
You need Node 22+ and a Personal Access Token from the md-log web app (the tokens section of settings; it's shown once). No token, no backend calls. Two ways to connect: the hosted HTTP endpoint at https://mcp.md-log.com/mcp, or local stdio through npx -y md-log-mcp.
What it helps with
- save an agent report as markdown at a chosen path
- upload screenshots alongside the text and get
asset://links back - append a new section to an existing report without rewriting it
- replace one exact fragment and leave the rest untouched
- roll a file back to an earlier version
Example prompts for the agent
- Save the error report to
my-project/2026-07-07-error-report.md - Append a conclusions section to the end of the report
- Replace the version line in the report with the current one
How to install md-log-mcp
Platform CLI
claude mcp add md-log-mcp -e MDLOG_API_BASE_URL=<MDLOG_API_BASE_URL> -e MDLOG_PAT=<MDLOG_PAT> -- npx -y md-log-mcpReplace <…> with your own values
Platform CLI
codex mcp add md-log-mcp --env MDLOG_API_BASE_URL=<MDLOG_API_BASE_URL> --env MDLOG_PAT=<MDLOG_PAT> -- npx -y md-log-mcpReplace <…> with your own values
Config file
{
"mcpServers": {
"md-log-mcp": {
"command": "npx",
"args": [
"-y",
"md-log-mcp"
],
"env": {
"MDLOG_API_BASE_URL": "<MDLOG_API_BASE_URL>",
"MDLOG_PAT": "<MDLOG_PAT>"
}
}
}
}Replace <…> with your own values
Config file
{
"servers": {
"md-log-mcp": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"md-log-mcp"
],
"env": {
"MDLOG_API_BASE_URL": "<MDLOG_API_BASE_URL>",
"MDLOG_PAT": "<MDLOG_PAT>"
}
}
}
}Replace <…> with your own values
Platform CLI
gemini mcp add -e MDLOG_API_BASE_URL=<MDLOG_API_BASE_URL> -e MDLOG_PAT=<MDLOG_PAT> md-log-mcp npx -y md-log-mcpReplace <…> with your own values
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about md-log-mcp
How do I install md-log-mcp in Claude Code?
Run claude mcp add md-log-mcp -e MDLOG_API_BASE_URL=<MDLOG_API_BASE_URL> -e MDLOG_PAT=<MDLOG_PAT> -- npx -y md-log-mcp. Fill in your base URL and token.
What does md-log-mcp do?
It stores your agent's markdown reports in md-log by path, creates missing folders, uploads images and keeps file versions.
What do I need to run md-log-mcp?
Node 22+ and a Personal Access Token from the md-log web app. The service itself is hosted, so there's nothing to run yourself.
