Archive

An entry cannot be deleted — and this is not a limitation of the interface but a decision: nobody will restore deleted knowledge, and the reason it was once written down can no longer be recalled.

What happens on archiving

The entry:

• stops getting into the index and the prompts — the model no longer sees it;
• moves from the “Active” tab to the “Archive” tab;
• keeps who removed it, when and for what reason.

Write the reason in plain words: “the rule was dropped”, “merged with the entry about the build”, “the module was decommissioned”. Half a year later, this is the only thing that will explain the decision.

Who sees the archive

Everyone with permission to read the knowledge base. The archive is not a dump and not a hidden part: it is the history of what the organisation once knew.

Restoring

The “Restore” button returns the entry to the active ones — it reaches the model again. The same permission is needed as for changing it.

The end of a task

Entries in the “task” scope go to the archive by themselves when the task ends — merged, rejected or stopped. First the pass after finished work moves into the project whatever will outlive the task, then the whole task scope is archived in one action with the reason “task #N completed”. Permanent task entries stay active — you set this flag precisely so that the entry is not removed. A task entry can be restored from the archive like any other.

What the AI archives

Only its own entries and only with a reason — when it has checked against the code and made sure that what was written is no longer true. It never archives your entries: at most it marks them “needs review” and explains what diverged. It does not touch permanent entries at all.