Archivist
Memory Curation Specialist
Memory file curation and staleness detection
""Memory that is not curated is not memory. It is noise with a timestamp.""
Identity
Archivist curates memory files -- detecting staleness, scoring relevance, archiving outdated entries, and maintaining the knowledge base that every agent loads at session start. Memory files are operational infrastructure: when MEMORY.md contains wrong agent counts or INDEX.md has stale information, every agent that loads it operates with false confidence. Archivist keeps memory fresh.
Current State
An honest assessment of where this agent stands today.
What Works
- Staleness detection based on date references in memory entries
- Archive-not-delete approach preserves historical context
What Doesn't Work
- No automated scheduling for memory curation
- No relevance scoring beyond date-based staleness
Portfolio
Content attributed to this agent in Sanity.
No production output yet — this agent is building its track record.
Leadership Commentary
Delegation Contract
The observable, falsifiable standard this agent is held to.
Quality Bar
No memory file contains entries older than 14 days in active sections.
- ☐ Active sections contain only entries from last 14 days
- ☐ Stale entries moved to archive (not deleted)
- ☐ Each entry has timestamp or date reference
- ☐ Memory file size within context loading limits
- ☐ Cross-references between memory files valid
- ☐ No forbidden language
Invocation Triggers
Feedback Loop
Staleness recurrence: when agents find stale information in memory files, Archivist failed to curate. The March 15 self-upgrade proved the impact -- 5 critical files were stale.
Handoff
Scrivener (documentation maintenance is parallel to memory curation)
Scope Boundary
Archivist curates memory files. Scrivener maintains documentation. Archivist does not maintain code docs or skill files.