Tool reference · Edit, history & reports
memory_forget
Applies the retention policy: archives old records that were never recalled and have low confidence, and purges records that stayed archived too long.
local stdio server writes destructive idempotent
Caution. With dry_run: false, archived and purged records lose their embeddings and stop appearing in search. Run memory_export first if you may need them.
When to use
- Memory has grown large and search is returning stale, never-used records.
- You want a preview of how many records the retention policy would affect (the default dry run).
Parameters
| Name | Type | Default | Description |
|---|---|---|---|
dry_run | boolean | true | If true, only show what would be affected |
Example
Arguments
{
"dry_run": true
} Result shape
{
"dry_run": true,
"would_archive": 14,
"would_purge": 3,
"archive_after_days": 180,
"purge_after_days": 365
} With dry_run: false the result is { applied: true, archived, purged }.
Values are illustrative; the keys follow the server's handler. MCP clients receive the result as JSON text content.
Server description
The description the server sends to your agent in tools/list, captured from the v14.7.0 source:
Apply retention policy: archive stale records (>180d, never recalled, low confidence), purge very old archived records (>365d). Keeps memory clean.