Mnemom — Trust Ratings for AI Agents

get_reputation

Read-onlyIdempotent

Look up an AI agent's published Trust Rating — Mnemom's portable reliability signal for autonomous software, computed from the agent's own verified activity record. Returns the rating plus the technical factors behind it. Free, public, read-only: ratings are public by default, unless the agent's owner has opted out.

Input Schema

TableJSON Schema

Name	Required	Description	Default
`agent_id`	Yes	Agent identifier (e.g. smolt-abc123)

Output Schema

TableJSON Schema

Name	Required	Description
`tier`	No
`grade`	Yes	AAA–D or NR.
`score`	Yes
`claimed`	No
`agent_id`	Yes
`trend_30d`	No
`agent_name`	No
`components`	Yes
`confidence`	Yes
`visibility`	Yes	Reputation-publication axis — whether this agent's Trust Rating is published. Every registered agent's reputation is `public` by accountability standard (the default; that is the whole point of a portable, verifiable rating); `private` is a rare owner opt-out that 403s the read to non-owners. This is DISTINCT from `Agent.public` (the identity-record visibility axis) — they share the word "public" but govern different things.
`computed_at`	No
`is_eligible`	Yes
`next_compute_at`	No	Next scheduled recompute — the 00/06/12/18 UTC cron slot strictly after `computed_at` (`floor(computed_at/6h)*6h + 6h`). Null when `computed_at` is null.
`checkpoint_count`	Yes
`a2a_trust_extension`	No	A2A trust extension for interop. Only present on `GET /reputation/{agent_id}` (not on batch/compare rows).
`checkpoint_accounting`	No	Structured breakdown of how checkpoints were counted toward the score. `analyzed` is the scoring population; `excluded` buckets are mutually exclusive and `analyzed + synthetic + insufficient_thinking + quarantined = total`. Null for legacy rows computed before this field existed.

Schema Changelog

Changes observed during successful MCP inspections.

Changed2 schema fields changedJul 27, 2026

addedOutput schema / $defs / ReputationScore / properties / checkpoint_accounting / properties / re_evaluated

Added value: +{
+  "description": "How many of the agent's checkpoints carry each kind of re-evaluation annotation (counted over `total`, not the analyzed subset). Observability only — these are not an exclusion bucket and do not enter the score arithmetic. Absent on scores computed before this breakdown existed.",
+  "properties": {
+    "corrected_clear": {
+      "description": "Checkpoints explicitly corrected to `clear` (`re_evaluation_metadata.corrected_verdict = 'clear'`).",
+      "type": "integer"
+    },
+    "corrected_non_clear": {
+      "description": "Checkpoints corrected to a NON-clear verdict — a cross-turn escalation or a reviewer reclassification to `review_needed`/`boundary_violation`. These count against the score per the correction, not as exonerations; a non-zero value is why this agent's score differs from the pre-MNE-2156 calculation.",
+      "type": "integer"
+    },
+    "resolved_no_correction": {
+      "description": "Checkpoints re-evaluated with NO recorded correction — the trust-recovery convention, scored as `clear`.",
+      "type": "integer"
+    }
+  },
+  "required": [
+    "corrected_clear",
+    "corrected_non_clear",
+    "resolved_no_correction"
+  ],
+  "type": "object"
+}

addedOutput schema / properties / checkpoint_accounting / properties / re_evaluated

Added value: +{
+  "description": "How many of the agent's checkpoints carry each kind of re-evaluation annotation (counted over `total`, not the analyzed subset). Observability only — these are not an exclusion bucket and do not enter the score arithmetic. Absent on scores computed before this breakdown existed.",
+  "properties": {
+    "corrected_clear": {
+      "description": "Checkpoints explicitly corrected to `clear` (`re_evaluation_metadata.corrected_verdict = 'clear'`).",
+      "type": "integer"
+    },
+    "corrected_non_clear": {
+      "description": "Checkpoints corrected to a NON-clear verdict — a cross-turn escalation or a reviewer reclassification to `review_needed`/`boundary_violation`. These count against the score per the correction, not as exonerations; a non-zero value is why this agent's score differs from the pre-MNE-2156 calculation.",
+      "type": "integer"
+    },
+    "resolved_no_correction": {
+      "description": "Checkpoints re-evaluated with NO recorded correction — the trust-recovery convention, scored as `clear`.",
+      "type": "integer"
+    }
+  },
+  "required": [
+    "corrected_clear",
+    "corrected_non_clear",
+    "resolved_no_correction"
+  ],
+  "type": "object"
+}

First observedJun 12, 2026

TDQS

A3.9/5.0

Behavior4/5

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

Annotations already declare readOnlyHint, idempotentHint, non-destructive and closed-world, so safety is covered. The description adds genuinely new behavioral context: the operation is free, ratings are public by default, and an owner opt-out can mean no published rating exists — a failure mode the annotations do not convey. It stops short of stating auth requirements or the exact response shape.

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?

Two tight sentences, front-loaded with the action and result, followed by the access/privacy conditions. Every clause carries information; nothing is 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?

An output schema exists, so the description need not enumerate return fields, and it still flags that technical factors accompany the rating. For a single-parameter read-only lookup with full annotation coverage, nothing an agent needs to call it correctly is missing.

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?

Only one parameter (agent_id) and schema description coverage is 100%, so the schema already documents it fully (with an example like smolt-abc123). The description adds no syntax, format, or resolution guidance for agent_id, so the baseline of 3 applies.

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?

Specific verb+resource: 'Look up an AI agent's published Trust Rating,' with a clarifying definition of what the rating is (a portable reliability signal computed from the agent's verified activity record). It also states what comes back (the rating plus the technical factors), which distinguishes it somewhat from a badge tool, but it never names or contrasts verify_reputation, search_reputation_directory, or get_reputation_badge despite those being close siblings.

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 gives contextual conditions — free, public, read-only, and that ratings exist by default unless the owner opted out — which implies when this lookup is appropriate. However, it offers no explicit when-to-use versus the several reputation-related siblings (verify_reputation, search_reputation_directory, get_reputation_badge, scan_trust), leaving the routing decision to inference.

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

All Mnemom — Trust Ratings for AI Agents 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.