Related Servers

Most tools occupy clearly separate niches, and the descriptions actively police the boundaries (work_plan vs shiftwork_plan, build_identity vs bantamkit_status vs shiftwork_status, memory_compact vs memory_dream each say when to use the other). The remaining ambiguity is real but narrow: the read-only 'plan' pair and the three status/diagnostic tools share vocabulary and could be mispicked without reading the full text. No two tools actually do the same thing.

Everything is snake_case with sensible namespace prefixes (shiftwork_*, memory_*, work_plan), which makes the surface easy to scan. The deviation is word order: memory_save/token_ledger/skill_audit are noun-first while validate_json/build_identity are verb-first, and bantamkit_status is server-prefixed where shiftwork_status is domain-prefixed.

Sixteen tools is on the heavy side, but this is a multi-surface server (shift-work orchestration 6, memory 5, diagnostics 2, plus ledger, audit, validation), and each tool maps to an operation no sibling covers. Nothing looks like filler, though the set is larger than a single-purpose server should ever be.

Shift-work has a full lifecycle (draft, plan, clock-in, clock-out, status) and memory has save/recall/compact/dream/queries, so the main domains are closed loops. Gaps exist but are documented as deliberate (no tool edits a checkpoint in progress, no tool deletes a memory fact, both deferred to operator commands), so an agent rarely hits a dead end it wasn't warned about.