Skip to main content
Glama

Connections

by icu.connections
OverviewDiscussionsAdmin

Hydra Core - the workspace's decision flowchart for AI chats

connections_core

Use this when an AI chat starts a task and needs to know how to do it: whether to do it itself or delegate, how big it is, which manager, worker or tool it goes to, and the brief to hand them. One flowchart per workspace, kept in the CrAInium and edited live. action run (default) walks it: send the answers you have (by question id) and facts such as { message }; it answers the next question to answer (status ask) or the outcome with the path taken, the people, the tools and a brief (status done). Most tasks end at the first questions with outcome myself: go do it. action head lists the flowchart's quick exits: the questions a chat answers itself before it acts, and which answers end there, so those need no call. action find lists the CrAInium docs and staff in this workspace that match a query or a message, as addresses to read, never their text. action get returns the flow (format outline for a readable view); edit changes it with ops (add_node, update_node, remove_node, add_edge, update_edge, remove_edge, set) and save replaces the whole graph, both with base_version and a why; revert brings an older version back; history lists versions; log lists recent runs; brief gives a chat back its last delegate brief after a compact; turns stores one ledger row per chat turn from a PC's hook: whether the chat walked the core and whether it delegated. Every version is kept.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
toNorevert: the version to bring back.
opsNoedit: [{ op: 'add_node', node: {id, type, ...} }, { op: 'update_node', id, set: {...} }, { op: 'remove_node', id }, { op: 'add_edge', edge: {from, to, when} }, { op: 'update_edge', id, set }, { op: 'remove_edge', id }, { op: 'set', set: { title, start, notes } }]. Node types: ask (text, answers), check (test: { fact, op, value }), do (say, use), person (name, role, mode consult|hire|lead, brief), end (outcome, say).
whyNoedit/save/revert: what changed and why, kept with the version.
chatNorun/log/brief: this chat's key, so its runs and brief can be found again.
flowNoWhich flowchart (default main).
kindNolog: which core_events rows (default run).
rowsNoturns: one row per chat turn, 1 to 200: { uid (the turn's key, 1-80 letters, digits, . _ : or -), chat, at (ISO time), called, delegated, version, pc }. called true means the chat walked the flowchart.
factsNorun: facts the flowchart's checks read, e.g. { message: '<the request>' }. find: { message } in place of query.
graphNosave: the whole flowchart { title, start, notes, nodes, edges }.
limitNohistory/log: how many (default 50, at most 500); find: docs to list (default 8, at most 20).
queryNofind: the words to match against the workspace's docs and staff (default facts.message).
actionNorun (default) walks the flowchart; head reads its quick exits; find reads matching docs and staff; get reads it; edit/save/revert change it; history, log, brief and turns read or store the record.
formatNoget: graph (default) or a readable outline.
answersNorun: answers so far, by question id, e.g. { need: 'no', mid: 'no', who: 'delegate', tier: 3 }.
companyNoThe workspace (project id) when this account belongs to more than one.
versionNoget: an older version to read.
folderPathNoA local folder bound to a workspace; picks that workspace's flowchart.
base_versionNoedit/save/revert: the version you read (0 for a new flowchart). A newer save is answered stale_version.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed8 schema fields changed
    • changedInput schema / properties / action / description
      Previous value: -"run (default) walks the flowchart; get reads it; edit/save/revert change it; history, log and brief read the record."New value: +"run (default) walks the flowchart; head reads its quick exits; find reads matching docs and staff; get reads it; edit/save/revert change it; history, log, brief and turns read or store the record."
    • changedInput schema / properties / action / enum
      Previous value: -[
-  "run",
-  "get",
-  "save",
-  "edit",
-  "revert",
-  "history",
-  "log",
-  "brief"
-]New value: +[
+  "run",
+  "head",
+  "find",
+  "turns",
+  "get",
+  "save",
+  "edit",
+  "revert",
+  "history",
+  "log",
+  "brief"
+]
    • changedInput schema / properties / base_version / description
      Previous value: -"edit/save/revert: the version you read (0 for a new flowchart). A newer save refuses with stale_version."New value: +"edit/save/revert: the version you read (0 for a new flowchart). A newer save is answered stale_version."
    • changedInput schema / properties / facts / description
      Previous value: -"run: facts the flowchart's checks read, e.g. { message: '<the request>' }."New value: +"run: facts the flowchart's checks read, e.g. { message: '<the request>' }. find: { message } in place of query."
    • addedInput schema / properties / kind
      Added value: +{
+  "description": "log: which core_events rows (default run).",
+  "enum": [
+    "run",
+    "turn",
+    "all"
+  ],
+  "type": "string"
+}
    • changedInput schema / properties / limit / description
      Previous value: -"history/log: how many (default 50, at most 500)."New value: +"history/log: how many (default 50, at most 500); find: docs to list (default 8, at most 20)."
    • addedInput schema / properties / query
      Added value: +{
+  "description": "find: the words to match against the workspace's docs and staff (default facts.message).",
+  "type": "string"
+}
    • addedInput schema / properties / rows
      Added value: +{
+  "description": "turns: one row per chat turn, 1 to 200: { uid (the turn's key, 1-80 letters, digits, . _ : or -), chat, at (ISO time), called, delegated, version, pc }. called true means the chat walked the flowchart.",
+  "items": {
+    "type": "object"
+  },
+  "type": "array"
+}
  2. Added

TDQS

A3.9/5.0
Behavior4/5

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

With annotations present the bar is lower, and the description still adds real behavioral context: default action, the ask/done status protocol and return shape (outcome, path, people, tools, brief), versioning semantics ('Every version is kept'), the stale_version conflict response tied to base_version, and the list of edit ops. The one gap is that save 'replaces the whole graph' while destructiveHint is false — the versioning/revert story justifies that, so I do not treat it as a contradiction.

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

Conciseness3/5

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

Front-loaded with purpose and the default action, and most sentences do carry action-specific meaning. But 11 actions are compressed into a single unbroken paragraph of run-on clauses, and a bulleted per-action structure would have made the routing far easier to scan. Several clauses are also compressed to the point of ambiguity ('turns stores one ledger row per chat turn from a PC's hook').

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 tool with 18 parameters, nested objects, 11 actions and no output schema, the description covers the necessary ground: what each action does, the default, the conflict behavior, and the shape of responses for the main read actions. It is close to complete, though a couple of actions (turns input contract, find's return ordering) are only thinly described.

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, but the description goes further by partitioning the 18 parameters across the 11 actions (chat for run/log/brief, ops/why/base_version for edit/save/revert, facts/answers for run, rows for turns, query/limit for find). The one thing it does not add is the syntax of the answers/facts objects beyond the schema's own examples.

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 description states a concrete resource (the workspace's decision flowchart / CrAInium flow) and enumerates all 11 actions with distinct verbs (walk, list, find, get, edit, save, revert, history, log, brief, store). It is clearly distinguishable from siblings such as connections_execute or connections_notes_search. However, the core identity is embedded in a dense wall of text rather than front-loaded as a single clear statement.

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?

It gives a strong when-to-use trigger ('when an AI chat starts a task and needs to know how to do it'), states that run is the default, and even implies exclusions ('head lists the flowchart's quick exits... so those need no call'). It does not compare itself to any sibling tool or state when this tool should be avoided in favor of another, so it stops short of a 5.

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

All Connections 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