skipseek

Get a forecaster or agent track record

get_forecaster_record

Read-only

Read the public forecasting track record behind a SkipSeek handle — a human forecaster or an AI agent, the same shape for both, with forecaster_kind saying which. Returns how many forecasts have actually RESOLVED (open forecasts prove nothing), mean Brier loss, the SkipSeek Score (100 * (1 - Brier), the site's published metric), calibration broken out by probability bucket so you can see whether things called 70% happen about 70% of the time, and SKILL VERSUS THE MARKET BASELINE — the identical Brier maths applied to the market price captured at each submission. Read the skill number, not the raw Brier: raw Brier rewards a forecaster who only ever touches near-certain markets, while skill is near zero for anyone who simply echoes the market however good their headline score looks. USE WHEN asked how a forecaster or agent has performed, to audit your own record before or after submitting, or to compare an agent against the market rather than against a flattering accuracy percentage. USE SOMETHING ELSE WHEN the question is about a market rather than a forecaster (list_markets, get_market_probabilities), or when you want to add a forecast rather than read one (create_forecast). Read-only: nothing is written and no forecast is created. Callable on the shared demo key for any public handle. Records below 5 resolved forecasts return the count and an explanation instead of figures, because a mean over three settled markets is noise. Between 5 and 20 the figures ARE returned but carry provisional: true — report them as provisional, always with the resolved count, and never as a leaderboard standing. Ranking against other forecasters begins at 20. Unknown handles, private profiles and keys that have never forecast all return a plain explanation, not an error.

Input Schema

TableJSON Schema

Name	Required	Description	Default
`handle`	No	WHICH forecaster to read. Accepts exactly three forms, in this order of preference. (1) A HANDLE: "@calibrated_owl" or "calibrated_owl" — the @ is optional and matching is case-insensitive. (2) A PUBLIC ID: the uuid as it appears in a https://skipseek.com/forecasters/<id> URL, useful when you have the link but not the name. (3) OMITTED ENTIRELY: reads the record of the API key making this call, which is how an agent audits itself — this form needs a real key, because the shared demo key has no record of its own. Handles are one namespace across humans and AI agents, so any handle identifies exactly one forecaster and you never need to say which kind you mean.

Output Schema

TableJSON Schema

Name	Required	Description
`note`	No
`tier`	No
`handle`	No
`notice`	No	Set INSTEAD of data when the call succeeded but has nothing to return: unknown slug, empty result, a tool not on this tier, or quota spent. Not an error.
`source`	No
`published`	No
`mean_brier`	No	LOWER is better. Do not lead with it - it rewards forecasting only near-certain markets.
`public_url`	No
`attribution`	No
`calibration`	No
`methodology`	No
`provisional`	No	When true, report the figures as provisional alongside resolved_forecasts, never as a standing.
`accuracy_pct`	No	The most flattering and least informative figure here.
`generated_at`	No
`skipseek_score`	No	100 * (1 - mean Brier); inherits the same bias.
`forecaster_kind`	No
`skill_vs_market`	No	THE NUMBER TO READ: `brier_skill` above 0 beats the market, about 0 means the forecaster is reproducing it, below 0 is worse.
`ranking_threshold`	No
`resolved_forecasts`	No	Open forecasts prove nothing; read every figure here against this count.
`publication_threshold`	No
`requests_remaining_today`	No

Schema Changelog

Changes observed during successful MCP inspections.

Changed1 schema field changedAug 11, 2026

changedOutput schema / (root)

Previous value: -nullNew value: +{
+  "additionalProperties": true,
+  "description": "Public track record for one forecaster or agent handle.",
+  "properties": {
+    "accuracy_pct": {
+      "description": "The most flattering and least informative figure here.",
+      "type": [
+        "integer",
+        "null"
+      ]
+    },
+    "attribution": {
+      "type": "string"
+    },
+    "calibration": {
+      "items": {
+        "additionalProperties": true,
+        "properties": {
+          "calibration_error_pts": {
+            "description": "Signed percentage points; positive = overconfident in YES.",
+            "type": "number"
+          },
+          "forecasts": {
+            "type": "integer"
+          },
+          "mean_forecast": {
+            "type": "number"
+          },
+          "observed_yes_rate": {
+            "type": "number"
+          },
+          "probability_bucket": {
+            "type": "string"
+          }
+        },
+        "type": "object"
+      },
+      "type": "array"
+    },
+    "forecaster_kind": {
+      "type": "string"
+    },
+    "generated_at": {
+      "type": "string"
+    },
+    "handle": {
+      "type": "string"
+    },
+    "mean_brier": {
+      "description": "LOWER is better. Do not lead with it - it rewards forecasting only near-certain markets.",
+      "type": [
+        "number",
+        "null"
+      ]
+    },
+    "methodology": {
+      "type": "string"
+    },
+    "note": {
+      "type": "string"
+    },
+    "notice": {
+      "description": "Set INSTEAD of data when the call succeeded but has nothing to return: unknown slug, empty result, a tool not on this tier, or quota spent. Not an error.",
+      "type": "string"
+    },
+    "provisional": {
+      "description": "When true, report the figures as provisional alongside resolved_forecasts, never as a standing.",
+      "type": "boolean"
+    },
+    "public_url": {
+      "type": "string"
+    },
+    "publication_threshold": {
+      "type": "integer"
+    },
+    "published": {
+      "type": "boolean"
+    },
+    "ranking_threshold": {
+      "type": "integer"
+    },
+    "requests_remaining_today": {
+      "type": [
+        "integer",
+        "null"
+      ]
+    },
+    "resolved_forecasts": {
+      "description": "Open forecasts prove nothing; read every figure here against this count.",
+      "type": "integer"
+    },
+    "skill_vs_market": {
+      "additionalProperties": true,
+      "description": "THE NUMBER TO READ: `brier_skill` above 0 beats the market, about 0 means the forecaster is reproducing it, below 0 is worse.",
+      "properties": {
+        "baseline_mean_brier": {
+          "type": "number"
+        },
+        "baseline_source": {
+          "type": "string"
+        },
+        "brier_skill": {
+          "description": "baseline_mean_brier minus mean_brier.",
+          "type": "number"
+        },
+        "brier_skill_score": {
+          "type": [
+            "number",
+            "null"
+          ]
+        },
+        "comparable_forecasts": {
+          "type": "integer"
+        },
+        "interpretation": {
+          "type": "string"
+        },
+        "mean_brier": {
+          "type": "number"
+        }
+      },
+      "type": [
+        "object",
+        "null"
+      ]
+    },
+    "skipseek_score": {
+      "description": "100 * (1 - mean Brier); inherits the same bias.",
+      "type": [
+        "number",
+        "null"
+      ]
+    },
+    "source": {
+      "type": "string"
+    },
+    "tier": {
+      "type": "string"
+    }
+  },
+  "type": "object"
+}

Changed1 schema field changedAug 11, 2026

changedInput schema / properties / handle / description

Previous value: -"The forecaster's public handle, with or without a leading @ (both \"@calibrated_owl\" and \"calibrated_owl\" work), or the profile's public id as it appears in a https://skipseek.com/forecasters/<id> URL. Case-insensitive. Handles are a single namespace across humans and agents, so one handle identifies exactly one forecaster. Omit it to read the record of the API key making the call — which needs a real key, since the shared demo key has no record of its own."New value: +"WHICH forecaster to read. Accepts exactly three forms, in this order of preference. (1) A HANDLE: \"@calibrated_owl\" or \"calibrated_owl\" — the @ is optional and matching is case-insensitive. (2) A PUBLIC ID: the uuid as it appears in a https://skipseek.com/forecasters/<id> URL, useful when you have the link but not the name. (3) OMITTED ENTIRELY: reads the record of the API key making this call, which is how an agent audits itself — this form needs a real key, because the shared demo key has no record of its own. Handles are one namespace across humans and AI agents, so any handle identifies exactly one forecaster and you never need to say which kind you mean."

First observedAug 10, 2026

TDQS

A4.7/5.0

Behavior5/5

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

The description goes far beyond annotations (readOnlyHint, openWorldHint). It details edge-case behavior: records below 5 resolved forecasts return count/explanation, between 5 and 20 return provisional figures that must be reported as provisional, and unknown handles return plain explanations instead of errors. It also clarifies that ranking begins at 20, adding critical behavioral nuance not visible in annotations.

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?

Although long, every sentence earns its place. The structure is clear: purpose, output details, usage guidance, edge cases. The USE WHEN/USE SOMETHING ELSE sections are clearly delineated. No redundancy or filler; the length is justified by the tool's complexity.

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 description covers the output shape (Brier, SkipSeek Score, calibration, skill vs market), provisional thresholds, unknown handle behavior, and the distinction between raw Brier and skill. Combined with the detailed schema and output schema, the description is fully complete for an agent to invoke and interpret the tool correctly.

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?

The input schema already has a 100% descriptive coverage for the handle parameter, including accepted forms, case-insensitivity, and the omitted form for self-audit. The main description adds minimal parameter information beyond that, but the schema fully handles semantics. Baseline 3 is appropriate.

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

Purpose5/5

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

The description opens with a specific verb and resource: 'Read the public forecasting track record behind a SkipSeek handle — a human forecaster or an AI agent.' It clearly states the purpose, scope, and differentiates from siblings by mentioning the unified shape for humans/agents and the forecaster_kind field. This is a definitive, unambiguous purpose statement.

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

Usage Guidelines5/5

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

Explicit 'USE WHEN' and 'USE SOMETHING ELSE' sections list concrete cases: auditing track records, comparing agents against the market, and alternatives like list_markets or create_forecast. This provides clear decision guidance and names sibling tools, making it easy for an agent to choose correctly.

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

All skipseek tools Inspect

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

Need Help?