Skip to main content
Glama

TeamShared

by io.github.xhad
OverviewDiscussionsAdmin

memory_recall

Hybrid recall across memory pillars within the caller's org.

Default scope searches durable pillars only (semantic, episodic, procedural, skill, strategic, work). Pass scope=["working"] to include this chat's open session turns. Shared brain on durable pillars: pass agent="cursor" only to narrow semantic/episodic. For entity/competitor questions use a short keyword anchor in query (e.g. "mex") plus repo / github. Use explain=true; prefer hits with matched_keyword: true. Default recall is current truth (superseded/merged rows are omitted); pass include_superseded=true for the replacement chain. When namespace= is set, semantic/episodic hits outside that container are excluded. Hits include a small provenance citation block (id, kind, timestamps, optional title, optional source) — metadata only, never page HTML. Result includes org + bound.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
kNoMax records to return
repoNoWorkspace slug of your current repo. When set, durable memories tagged for this repo are boosted (ranked higher); nothing is hidden — cross-repo and un-scoped memories still appear. Pass your workspace slug when recalling for code/repo-specific work.
agentNoOptional filter — restrict semantic/episodic results to this agent's writes. Default (None) is the shared brain: every agent's durable memories in the org are visible. Working memory is always scoped to the caller regardless.
queryYesNatural-language query
scopeNoPillars to search. Default (null): durable pillars only (semantic, episodic, procedural, skill, strategic, work) — working is omitted. Add scope=['working'] when you need this chat's open session turns.
githubNoGitHub repository as owner/repo. Boosts memories tagged github:<owner>/<repo> (portable across checkout paths).
explainNoWhen true, include per-record retrieval attribution in metadata.ranking (base RRF, every factor including bounded feedback trust from memory_feedback votes, relevance floor verdict, diversity penalty; its score equals the hit's score), plus read-only feedback_score when votes exist
filtersNoOptional AND filter. Keys (all optional): kind (fact|preference|event|note|outreach|skill|procedure), pillar (semantic|episodic|procedural|skill|strategic|work), subject (exact, case-insensitive), tags (list; record must include every tag), since / until (ISO datetimes). Example: {"kind": "fact", "subject": "HolderBrief", "tags": ["decision"]}. Nested AND/OR/NOT is not supported — omit a key instead. Empty result retries once without the filter (filter_relaxed=true on the result). Not for named playbook/skill/entity — use get-by-name tools.
verboseNoWhen false, truncate record content and omit metadata
namespaceNoOrg-allowlisted container. When set, semantic/episodic hits outside this slug are excluded (not merely demoted). Omit to keep shared-org recall. repo=/github= still boost. Unknown slugs fail. Example: github:acme/app or project:alpha.
time_rangeNoOptional time bounds for episodic/working hits
deadline_secondsNoOptional wall-clock budget for this call. Finished pillars are returned with errors_by_pillar timeout codes and degraded=true instead of letting the MCP client kill the whole call. Omit to use the server TEAMSHARED_RECALL_DEADLINE_SECONDS default (unset = no cutoff).
include_supersededNoWhen true, also return superseded/merged history. Default is current truth only.

Output Schema

TableJSON Schema
NameRequiredDescriptionDefault

No arguments

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed1 schema field changed
    • changedInput schema / properties / explain / description
      Previous value: -"When true, include per-record retrieval attribution in metadata.ranking (base RRF, every factor, relevance floor verdict, diversity penalty; its score equals the hit's score), plus read-only feedback_score when votes exist"New value: +"When true, include per-record retrieval attribution in metadata.ranking (base RRF, every factor including bounded feedback trust from memory_feedback votes, relevance floor verdict, diversity penalty; its score equals the hit's score), plus read-only feedback_score when votes exist"
  2. Changed1 schema field changed
    • changedInput schema / properties / explain / description
      Previous value: -"When true, include per-record retrieval attribution in metadata, plus read-only feedback_score when votes exist"New value: +"When true, include per-record retrieval attribution in metadata.ranking (base RRF, every factor, relevance floor verdict, diversity penalty; its score equals the hit's score), plus read-only feedback_score when votes exist"
  3. Changed1 schema field changed
    • addedInput schema / properties / namespace
      Added value: +{
+  "anyOf": [
+    {
+      "type": "string"
+    },
+    {
+      "type": "null"
+    }
+  ],
+  "default": null,
+  "description": "Org-allowlisted container. When set, semantic/episodic hits outside this slug are excluded (not merely demoted). Omit to keep shared-org recall. repo=/github= still boost. Unknown slugs fail. Example: github:acme/app or project:alpha."
+}
  4. Changed1 schema field changed
    • changedInput schema / properties / explain / description
      Previous value: -"When true, include per-record retrieval attribution in metadata"New value: +"When true, include per-record retrieval attribution in metadata, plus read-only feedback_score when votes exist"
  5. Changed1 schema field changed
    • addedInput schema / properties / deadline_seconds
      Added value: +{
+  "anyOf": [
+    {
+      "exclusiveMinimum": 0,
+      "type": "number"
+    },
+    {
+      "type": "null"
+    }
+  ],
+  "default": null,
+  "description": "Optional wall-clock budget for this call. Finished pillars are returned with errors_by_pillar timeout codes and degraded=true instead of letting the MCP client kill the whole call. Omit to use the server TEAMSHARED_RECALL_DEADLINE_SECONDS default (unset = no cutoff)."
+}
  6. Changed1 schema field changed
    • addedInput schema / properties / filters
      Added value: +{
+  "anyOf": [
+    {
+      "additionalProperties": true,
+      "type": "object"
+    },
+    {
+      "type": "null"
+    }
+  ],
+  "default": null,
+  "description": "Optional AND filter. Keys (all optional): kind (fact|preference|event|note|outreach|skill|procedure), pillar (semantic|episodic|procedural|skill|strategic|work), subject (exact, case-insensitive), tags (list; record must include every tag), since / until (ISO datetimes). Example: {\"kind\": \"fact\", \"subject\": \"HolderBrief\", \"tags\": [\"decision\"]}. Nested AND/OR/NOT is not supported — omit a key instead. Empty result retries once without the filter (filter_relaxed=true on the result). Not for named playbook/skill/entity — use get-by-name tools."
+}
  7. Changed1 schema field changed
    • addedInput schema / properties / include_superseded
      Added value: +{
+  "default": false,
+  "description": "When true, also return superseded/merged history. Default is current truth only.",
+  "type": "boolean"
+}
  8. First observed

TDQS

A4.2/5.0
Behavior4/5

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

No annotations are present, so the description carries the full burden. It discloses defaults (durable-only scope, current-truth semantics, shared-brain visibility), scoping exclusions (namespace), and the provenance citation block (metadata only, never page HTML), plus result fields (org + bound). It doesn't cover failure modes or rate limits, but for a read-style recall it is substantially transparent.

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

Conciseness5/5

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

Around five dense sentences front-load the purpose, then alternate pithy parameter-guidance sentences. Each clause earns its place and there is minimal fluff. Some default-behavior statements mirror schema descriptions, but given the 13-param complexity the repetition is acceptable guidance, not padding.

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

Completeness5/5

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

The package is strong: detailed per-param descriptions cover the rest (filters, deadline_seconds, k, verbose) and an output schema exists. The description covers the non-obvious semantics (current-truth default, provenance shape, namespace exclusion, working-memory scope) that an agent must know to invoke it correctly. Nothing critical for selection or invocation is missing.

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

Parameters4/5

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

Schema coverage is 100%, so the baseline is 3. The description adds strategy beyond the schema: entity/competitor queries warrant a short keyword anchor with repo/github, explain=true should be used, and matched_keyword hits are preferred. It also clarifies shared-brain vs agent='cursor' behavior. This pushes it a step above baseline.

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?

The opening line 'Hybrid recall across memory pillars within the caller's org' gives a specific verb (recall), a clear resource (memory pillars), and an org-scoped target. The follow-up enumerates the six durable pillars and the working-memory option, which distinguishes it from other memory_* tools without naming them. It stops short of a crisp one-line contrast with siblings such as memory_entity_view or memory_think, so 4.

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

Usage Guidelines4/5

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

The description provides clear when-to-use tips: pass scope=['working'] for this chat's open session turns, agent='cursor' to narrow semantic/episodic, include_superseded for the replacement chain, and namespace to constrain the container. It also gives query-shaping advice (short keyword anchor + repo/github for entity/competitor questions, explain=true, prefer matched_keyword). It does not explicitly say when not to use this tool in favor of a sibling, so no exclusion statement is present.

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

All TeamShared toolsInspect
Try in Browser
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources