Overview
contox_compact summarizes all entries in a journal-like context (bugs, implementation journal, sessions) into a single content block. This reduces the token count of the context while preserving the key information from all entries.
Use this tool when a journal context has accumulated many entries over time and is consuming too many tokens in the brain document.
Parameters
| Name | Type | Required | Default | Description |
|---|---|---|---|---|
id | string | Yes | — | The context ID to compact |
Return value
Returns a summary of the compaction operation:
| Field | Type | Description |
|---|---|---|
entriesCount | number | Number of entries that were processed |
summaryLength | number | Length of the resulting summary in characters |
message | string | Status message |
Usage examples
Compacting a bugs journal
User: The bugs context has too many entries, compact it
Claude calls: contox_list_contexts()
// Finds bugs context with id "ctx_bugs123"
Claude calls: contox_compact({ id: "ctx_bugs123" })
Response:
Compaction completed for context ctx_bugs123:
- Entries processed: 47
- Summary length: 2340 chars
- All entries summarized into context content
Notes
- Compaction is a one-way operation. Individual entries are summarized into the context's content field, and the original entries are consolidated.
- The summary preserves key facts, dates, and references from the original entries.
- After compaction, new entries can still be added to the context via
contox_add_entry. - Consider compacting when a context's token count grows large (visible via
contox_list_contexts).