Logs

A log is the chronological history of what was done: deployments, changes, incidents, migrations, tests. Each action is a new line. Lines cannot be deleted, and an AI connected through MCP can only append: it cannot edit or delete earlier ones.

That makes them useful as an audit trail of what your agents have done, and as a memory of what happened for the next session, whichever AI it is.

Rules

• Append-only. There is no MCP tool to edit or delete lines.
• Only you correct. From the web you can fix a line if you made a mistake.
• A log cannot be deleted.
• Self-contained lines. What was done, where, with what result and, if relevant, links to the notes or environments involved.
• They can be organized in folders and shared.

Notes vs logs

	Note	Log
Answers	How does it work? How is it done?	What happened? When?
Rewritten	Yes, reflects the current state	No, it only grows
Example	“How the API is deployed”	“Nov 10: v1.24.0 deployed, migration OK”

Rule of thumb: if you are about to write something dated, such as “today I did X”, or to add entries to a “History” section, it goes in a log. If how something works changes, update the note and add a line to the log.

How an AI uses them

Through the MCP server the AI finds the log with logs_list, creates it with log_create if it does not exist, and appends each action with log_append. At the start of a new session it reads the latest lines with log_get to know what happened before.