Skip to main content
Glama

Council of AI GSPC

by io.github.CSOAI-ORG
OverviewDiscussionsAdmin

GSPC Route: decide a path under your policy (free, decide-only)

route
Read-onlyIdempotent

GSPC Route (free, decide-only). Applies YOUR policy and the fixed GSPC floor to a candidate set and returns the chosen candidate, the basis of the choice and an unsigned route record (csoai.evidence-event/0.1, profile csoai.route-evidence/0.1). Candidates are the GSPC MCP tools by default, or ones you declare: your models, your local GPU, MCP tools, A2A agents, x402 resources. Policy is Cedar: presets read-only, local-only, eu-only, no-unmeasured, plus forbid_providers and allow_kinds; anything the router does not understand grants nothing. For the quality axis you name it reads the live board (https://councilof.ai/api/gspc). With no candidates declared, the GSPC tool whose purpose matches the task is chosen (basis task_match); when no tool's purpose matches, the answer is UNTESTED and nothing is chosen, never a pick by name. For candidates you declare, routing is not ranking: a choice is called a separated leader only when the board separated that comparison; otherwise it is TIE or UNTESTED and your tie_break decides. A candidate the board holds no number for is UNTESTED, never 0. Sponsor, bid and placement fields are ignored. The task text is hashed, never stored. Nothing is executed, called or charged; mode execute answers NOT_ENABLED. Measurement, not certification.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
modeNodecide (default). execute is not enabled and answers NOT_ENABLED.
taskNoThe task. Hashed (sha256) and never stored or echoed.
policyNopresets [read-only, local-only, eu-only, no-unmeasured], forbid_providers [..], allow_kinds [..], confirm_destructive, caller_wallet. Unknown keys grant nothing.
objectiveNo
candidatesNoOmit to route among the GSPC MCP tools. A candidate field the router does not know makes that candidate UNCHECKABLE (never permitted); sponsor/bid/placement fields are ignored.
data_classNoDefault public. A non-public class is routed only to candidates that declare it in data_class_allowed.
needs_writeNo
task_sha256NoAlternative to task: its sha256.

Output Schema

TableJSON Schema
NameRequiredDescriptionDefault
modeNo
noteNo
labelNo
stateYes
chosenNo
errorsNo
recordNo
signedNo
previewNo
forbiddenNo
permittedNo
consideredNo
separationNo
http_statusNo

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed1 schema field changed
    • changedOutput schema / properties / state / enum
      Previous value: -[
-  "ROUTED",
-  "NO_PERMITTED_CANDIDATE",
-  "BAD_ARGUMENTS",
-  "NOT_ENABLED"
-]New value: +[
+  "ROUTED",
+  "NO_PERMITTED_CANDIDATE",
+  "UNTESTED",
+  "BAD_ARGUMENTS",
+  "NOT_ENABLED"
+]
  2. Added

TDQS

A4.6/5.0
Behavior5/5

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

Annotations already declare readOnly=true/idempotent=true/destructive=false, and the description reinforces rather than contradicts these. It adds substantial beyond-annotation context: task text is hashed and never stored, nothing is executed/called/charged, unknown policy keys grant nothing, unmeasured candidates are UNTESTED (never 0), sponsor/bid/placement are ignored, and separation-of-leader semantics for declared candidates.

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?

It is a dense block but front-loads the core purpose and each sentence carries operational meaning (UNTESTED rules, tie_break, hashing, NOT_ENABLED). Some mild repetition of 'free, decide-only' from the title, and the run-on paragraph could be broken up, but waste is low for a tool of this 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?

An output schema exists so return values need no prose. For an 8-parameter tool with nested policy/objective/candidate objects and an open-world routing decision, the description covers the decision rules, failure/degenerate outcomes, and privacy behavior an agent needs to call it 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?

With ~75% schema coverage the baseline is 3, but the description adds real meaning: it enumerates policy presets, explains tie_break ordering, notes the quality axis reads the live board, describes data_class routing, and warns that an unknown candidate field makes it UNCHECKABLE while sponsor fields are ignored. Only the objective weights values (cost/latency/quality) go unexplained.

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?

States a specific action (applies YOUR policy + the fixed GSPC floor to a candidate set) and a specific return (the chosen candidate, the basis, and an unsigned route record). This is clearly distinguishable from read-only siblings like board_totals, get_axis, and the verify_* tools. An agent can tell what this does without opening the schema.

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?

Explains when it applies: candidates default to GSPC MCP tools or ones you declare; with none declared the tool matching the task purpose is chosen; when nothing matches the answer is UNTESTED. It also flags that mode=execute answers NOT_ENABLED, steering callers to decide. It never names a sibling as an alternative, but the usage conditions are concrete and unambiguous.

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

All Council of AI GSPC 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