Skip to main content
Glama

Tomorrow Lives Engines

by com.tomorrowlives
OverviewDiscussionsAdmin

Analyse pull sessions

pull_analyse

Analyses the sessions of one condition (one session, a list, or {"sessions": [...]}) together, and each alone, and returns the report: its label (significant; null, with the smallest effect these sessions could have found; unlabelled, with what is missing; or not testable) and the sentence behind it, the mean signed shift against three surrogate families, the placebos, the four cells (deliberate, programmed, luck and sync), intended against unintended changes, recovery, locking, and the report's hash. prereg registers the run; replication analyses a second set under it, beside the first; measure is phase (timing, the default) or state (the stance along the leader's declared direction, for two agents); surrogates (19 to 9,999, default 199) sets an unregistered run's draws a family, a registered run taking its pre-registration's; seed fixes them; a session of three or more generators is tested pair by pair, every ordered pair with its own ledger and the correction for their number, unless pairs is false; detail adds every pull record. It costs one unit for each 40,000 session-draws over a thousand events (a session's own reading counts as 200 draws), at least 1. A group costs its ordered pairs' worth. A call priced at more than 2 units is queued: the answer is 202 with a job id and its price, GET /api/agents/jobs/ (or the MCP tool job) says when it is done, and the whole report comes back as report.json beside a short verdict. A queued call is refused on the free plan with 402: its units buy only calls that answer at once, and queued jobs need a paid plan (a pass or the card plan). Sessions are not kept.

Input Schema

NameRequiredDescriptionDefault
seedNoA whole number that makes the run reproducible: the same seed and inputs give the same answer.
pairsNoTrue to analyse performer pairs as well.
detailNoTrue to return the fuller per-session detail.
preregNoThe pre-registration the sessions were run under, to check the analysis against it.
measureNoThe measure to analyse, by name.
sessionsYesOne session, a list of sessions, or {"sessions": [...]} of one condition, as JSON documents.
surrogatesNoHow many surrogate runs to draw for the null, 19 to 9999.
replicationNoSessions of a replication, analysed against the first set.

Output Schema

NameRequiredDescriptionDefault
msNoMilliseconds the run took.
jobYesThe job id; the job tool reads it.
verbNoThe tool that made it.
filesNoFiles made, each with a name, type, size and a signed URL.
unitsNoUnits this call cost.
resultNoThe engine's own answer, a JSON document.
statusNoqueued when the job runs in the background.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed11 schema fields changed
    • addedInput schema / properties / detail / description
      Added value: +"True to return the fuller per-session detail."
    • addedInput schema / properties / measure / description
      Added value: +"The measure to analyse, by name."
    • addedInput schema / properties / pairs / description
      Added value: +"True to analyse performer pairs as well."
    • addedInput schema / properties / prereg / description
      Added value: +"The pre-registration the sessions were run under, to check the analysis against it."
    • removedInput schema / properties / replication / $ref
      Removed value: -"#/properties/sessions"
    • addedInput schema / properties / replication / anyOf
      Added value: +[
+  {
+    "$ref": "#/properties/sessions/anyOf/0"
+  },
+  {
+    "$ref": "#/properties/sessions/anyOf/1"
+  }
+]
    • addedInput schema / properties / replication / description
      Added value: +"Sessions of a replication, analysed against the first set."
    • addedInput schema / properties / seed / description
      Added value: +"A whole number that makes the run reproducible: the same seed and inputs give the same answer."
    • addedInput schema / properties / sessions / description
      Added value: +"One session, a list of sessions, or {\"sessions\": [...]} of one condition, as JSON documents."
    • addedInput schema / properties / surrogates / description
      Added value: +"How many surrogate runs to draw for the null, 19 to 9999."
    • changedOutput schema / (root)
      Previous value: -nullNew value: +{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "additionalProperties": true,
+  "properties": {
+    "files": {
+      "description": "Files made, each with a name, type, size and a signed URL.",
+      "items": {
+        "additionalProperties": {},
+        "type": "object"
+      },
+      "type": "array"
+    },
+    "job": {
+      "description": "The job id; the job tool reads it.",
+      "type": "string"
+    },
+    "ms": {
+      "description": "Milliseconds the run took.",
+      "type": "number"
+    },
+    "result": {
+      "description": "The engine's own answer, a JSON document."
+    },
+    "status": {
+      "description": "queued when the job runs in the background.",
+      "type": "string"
+    },
+    "units": {
+      "description": "Units this call cost.",
+      "type": "number"
+    },
+    "verb": {
+      "description": "The tool that made it.",
+      "type": "string"
+    }
+  },
+  "required": [
+    "job"
+  ],
+  "type": "object"
+}
  2. First observed

TDQS

A3.5/5.0
Behavior4/5

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

Annotations mark it as non-read-only, non-idempotent, non-destructive, open-world false. The description adds substantial behavioral detail: cost model (40000 draws per unit), queuing threshold>2 units, 202/402 plan limits, and that sessions are not kept. This goes well beyond annotations. However, it doesn't clarify auth requirements or rate limits beyond the cost model.

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

Conciseness2/5

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

The description is a single, extremely long and dense paragraph with minimal structure. It is not front-loaded; key information about cost and queuing is buried in the middle. While concise in word count, it is hard to parse due to lack of organization.

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?

Given the complexity (8 parameters, output schema exists, nested logic), the description covers most behavioral aspects: cost, queuing, plan limits, and data retention. It might miss some edge cases but is fairly complete for an advanced tool.

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 coverage is 100%, so the schema already documents all parameters. The description adds some context about parameter behavior (e.g., surrogates range, measure default), but baselines at 3 when 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?

The description states a specific verb and resource: it analyses sessions of one condition and returns a detailed report. It is clear what the tool does, but it does not explicitly differentiate itself from siblings like pull_replay or pull_prereg, which likely also deal with session analysis.

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?

The description mentions related tools (prereg registers, replication analyses a second set), implying usage contexts, but it does not give explicit when-to-use vs. when-not-to-use guidance or name alternatives for this specific analysis task.

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

All Tomorrow Lives Engines 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