Skip to main content
Glama
LamantinAI

kaeru memory

by LamantinAI
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Rust
Hybrid

cite

Record an archival reference for long-term recall: capture external sources by URL or personas and entities without links, optionally linking each citation to an existing node in the same call.

Instructions

Record an archival reference. Two flavours: external source (pass url for papers / gists / dashboards) OR persona / entity (skip url for people, places, books without links). Both land in archival tier — long-term recall. Pass visibility=shared (in a team initiative) to capture and push to the cloud in one call. Pass link_to (with weight) to connect it to an existing node in this same call — an island is found only by exact name.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
urlNoOptional URL of the source. Skip for persona / entity records (a person, place, book without a link).
bodyYesOne-paragraph summary — what's at the link, or who this entity is.
nameYesShort, recallable name.
afterNoOptional: hold this back until `after` (YYYY-MM-DD), then surface it in `awake` as a debt. For something TRUE LATER, NOT NOW — a certificate that expires, a number to re-measure before quoting it again, work to pick up after the next release. Requires `for_days`.
cloudNoWhich cloud `visibility=shared` publishes to. Only consulted when sharing. Required when several clouds are configured — the push is refused rather than sent to a default you did not name.
layerNoOptional memory layer stamped at creation: `core`, `hot`, `warm`, `cold`, or `frozen`. Defaults to `warm`.
weightNoHow load-bearing the `link_to` edge is, 0..1. REQUIRED with `link_to`: it is the only signal knowledge chains route on, and there is no default because an unweighted graph makes every chain rank on noise. The capture still lands without it; the edge does not.
link_toNoOptional: the node this one connects to, by name or id — the edge is made in THIS call, while both ends are still in mind. Linking as a second step is the step nobody takes: one real vault reached 23 nodes and 0 edges with the nudge asking every time. Needs `weight`.
for_daysNoOptional: how many days the reminder keeps appearing once it surfaces. REQUIRED with `after`, and there is no default — state it deliberately, because "the certificate expires" and "re-measure this" want completely different windows. The window starts when the reminder is first actually seen, so it cannot expire while nobody is looking.
edge_typeNoEdge type for `link_to` — same closed vocabulary as `link` (`refers_to` by default, `causal`, `derived_from`, `contradicts`, `part_of`, `blocks`, `targets`, `supersedes`, `verifies`, `falsifies`, `temporal`, `consolidated_to`).
initiativeNo
visibilityNoOptional visibility. `shared` marks team knowledge and — in a `team` initiative with the secret guard clear — pushes it to the cloud in this one call. Defaults to `local` (stays private).

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv0.7.5

TDQS

A3.7/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations, the description carries the full burden and does so well: it discloses the archival tier, that shared visibility pushes to cloud in the same call, that link_to creates the edge inline, that islands are matched only by exact name, and that cloud is refused rather than defaulted. These are real behavioral traits beyond the schema text.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Purpose and the primary two-flavour branch are front-loaded, followed by the optional capabilities. Each sentence carries information, though the density makes it a slightly heavy read for a single description block.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 12-parameter, annotation-free tool with no output schema, the description covers the tricky interactions (weight required with link_to, for_days required with after, cloud conditional on visibility) that a schema alone would not make obvious. Coverage is strong, with only the sibling-routing dimension left thin.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 92%, so the schema already documents nearly every parameter. The description adds framing for the url flavour and reinforces weight/link_to coupling, but it is largely redundant with the structured parameter docs. Baseline 3 applies when the schema does the heavy lifting.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb+resource ('Record an archival reference') and cleanly splits the tool into two flavours (external source with url vs persona/entity without). The intended target is clear, though it does not explicitly distinguish itself from nearby siblings like jot, link, or attach.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines3/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Gives conditional guidance for the main branches (pass url vs skip; visibility=shared to push to cloud; after for 'true later, not now'), which is useful. However, it never frames when to choose cite over competing tools (jot, link, attach), leaving the sibling-selection decision to inference.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Deploy Server

Other Tools