Skip to main content
Glama

Palimpsest — censorship, China economy and model-eval observatory

by io.github.beepboop2025
OverviewDiscussionsAdmin

One signal's full reading

get_signal
Read-onlyIdempotent

Read one named signal. Permitted signals return their bounded latest payload; China-economic signals in the denied lineage closure return explicit restricted/unavailable rights metadata and no values. Call list_signals first to discover valid names. Use this for the AI-model-evaluation side too: 'eval-registry' returns the pre-registered, hash-chained eval ledger with its verified flag and Merkle root, 'gfi-transcripts' returns a bounded view of the complete GFI v2 response matrix, and 'refusal-drift' returns the current frontier-model refusal reading on the frozen benign probe set; read 'eval-assurance' before turning either into a validity claim, and 'eval-journal' for the evidence-bound explanation and source receipts, or 'eval-findings' for the current deterministic article edition. Distinct from gfw_reading, which merges the two Great Firewall layers into one combined view.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
nameYessignal name from list_signals, e.g. 'ooni-gfw', 'eval-registry', 'eval-assurance', 'eval-journal', 'eval-findings', 'gfi-transcripts' or 'refusal-drift'
max_rowsNocap on long row arrays (dataset, ranked, samples). Default 25 keeps a call small enough not to stall a tool loop; the generative-firewall-index dataset is 132 rows. Any cap applied is reported in the response with the true total, never silently.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed1 schema field changed
    • changedInput schema / properties / name / description
      Previous value: -"signal name from list_signals, e.g. 'ooni-gfw', 'eval-registry' or 'refusal-drift'"New value: +"signal name from list_signals, e.g. 'ooni-gfw', 'eval-registry', 'eval-assurance', 'eval-journal', 'eval-findings', 'gfi-transcripts' or 'refusal-drift'"
  2. Changed1 schema field changed
    • addedInput schema / properties / max_rows
      Added value: +{
+  "default": 25,
+  "description": "cap on long row arrays (dataset, ranked, samples). Default 25 keeps a call small enough not to stall a tool loop; the generative-firewall-index dataset is 132 rows. Any cap applied is reported in the response with the true total, never silently.",
+  "maximum": 500,
+  "minimum": 1,
+  "type": "integer"
+}
  3. Changed1 schema field changed
    • changedInput schema / properties / name / description
      Previous value: -"signal name from list_signals, e.g. 'ooni-gfw' or 'generative-firewall-index'"New value: +"signal name from list_signals, e.g. 'ooni-gfw', 'eval-registry' or 'refusal-drift'"
  4. Changed1 schema field changed
    • changedInput schema / properties / name / description
      Previous value: -"e.g. 'ooni-gfw'"New value: +"signal name from list_signals, e.g. 'ooni-gfw' or 'generative-firewall-index'"
  5. First observed

TDQS

A4.8/5.0
Behavior5/5

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

Beyond annotations (readOnly, idempotent), the description discloses that denied-lineage China signals return rights metadata instead of values, and that max_rows caps are reported with the true total rather than silently. This adds concrete behavioral context not present in the structured fields.

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

Conciseness4/5

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

The description is front-loaded with the core purpose and then expands. The long enumeration of eval signals is dense but each item adds specific value, so it earns its space. It is not succinct, but the complexity justifies the length.

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?

Even without an output schema, the description covers return behavior for permitted signals, denied signals, and each eval signal, plus the max_rows cap reporting and prerequisite list_signals call. An agent has everything needed to call the tool correctly.

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 baseline is 3, but the description adds meaningful semantics: it explains that max_rows caps are reported explicitly and that list_signals should be called first to get valid names. This goes beyond the schema's simple field descriptions.

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 'Read one named signal,' a specific verb+resource statement. It then clarifies the two output categories (bounded payload vs. restricted rights metadata) and explicitly distinguishes from gfw_reading, so an agent can tell them apart without opening schemas.

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?

It instructs to call list_signals first to discover valid names, and explicitly names gfw_reading as the alternative for a combined view. It also gives per-signal usage notes (e.g., read 'eval-assurance' before making validity claims), covering when and when not to use the tool.

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

All Palimpsest — censorship, China economy and model-eval observatory 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