Council of AI GSPC
Council of AI
独立的AI治理测量。我们测量、签署、再认证——任何人都可以核查。
Council of AI(CSOAI LTD,UK Companies House 16939677)是一个独立的AI行为测量机构。我们使用已发布且冻结的测量工具运行AI系统;将结果发布为经签署、可离线验证的测量凭证(Ed25519);并按固定周期重新测量,以确保证据保持时效。测量,而非认证——我们不认证、不出售评级、不提供整改服务,也不从我们排名的任何对象处收取费用。验证免费且无需登录。
🚀 快速开始
# Install via pip
pip install councilof_ai
# Or install via Smithery
npx -y @smithery/cli@latest install councilof-ai --client claudeRelated MCP server: Trust Chain MCP
📏 我们测量什么
13个GSPC测量轴——治理、安全、来源、连续性、合规性、开放性、机器一致性、关怀、跨现实、检测器互操作、第5条保障、群体、情感
基于冻结工具的确定性评分——无LLM裁判,无虚构分数
UNMEASURED是一等结果——差距会连同其n值和限制一起报告,从不隐藏,也从不计为失败
签署凭证——基于SHA-256的Ed25519,链式关联、时间锚定、可离线验证
实时轴数据:
GET https://councilof.ai/api/gspc(模式csoai.gspc-axes/0.3)
🧭 验证
每份已发布的测量都带有公钥和验证路径。可在以下位置核查任何凭证:
https://councilof.ai/verify — 免费,无需登录
📖 文档
⚖️ 我们从不做的事
❌ 认证AI系统或颁发合规徽章
❌ 出售评级、排名位置或成绩的提前知情权
❌ 以收费方式提供整改或推荐修复方案
❌ 从我们排名的任何对象处双向收取费用
🤝 Council of AI生态系统的一部分
界面 | 用途 |
测量机构——签署凭证、验证、记分板 | |
公共网站 | |
主权AI平台 |
📜 许可证
MIT © CSOAI-ORG
Available Tools
12 toolsart50_marking_evidenceAInspect
PAID (x402 or CSOAI LTD invoice). Article 50 marking-evidence pack via https://councilof.ai/api/art50/marking-evidence: is a machine-readable mark DETECTABLE in these bytes right now (C2PA manifest store, assertion hashes, hard binding, claim signature; IPTC digitalSourceType), beside the verbatim Art 50(2) excerpt hash and the Art 99(4) ceiling. Watermarks are UNCHECKABLE where no public detector exists and the pack says so. Point-in-time detection — never a conformity opinion, never a compliance word of any kind. preview=true is free and returns the same measurement unsigned. Without x_payment the tool returns the 402 challenge. If the route is not deployed on this origin the tool says NOT_DEPLOYED rather than inventing a result.
Inputs: url, preview, bytes_b64, x_payment, +1 more.
board_totalsAInspect
Live GSPC board totals from https://councilof.ai/api/gspc. Returns the slot count and the measured count as two labelled numbers WITH their kind — a slot is a declared position on the board, a measurement is a real run behind it; the two are never summed and never swapped — plus as_of dates for the board and for this fetch. We measure, never certify. If the board cannot be fetched the answer is a distinct UNREACHABLE state: no cached number is ever presented as live.
Inputs: no named parameters.
commission_cardAInspect
PAID (x402). Commission one signed card-v0 receipt (surface ras.commission) for a named subject on the frozen bank via https://councilof.ai/api/request-attestation. Re-serves every signed measurement card already on file for the subject; a payment never mints a MEASURED cell (fresh_run stays UNMEASURED until a published run exists). Without x_payment the tool returns the 402 challenge (accepts[] with asset, amount, payTo) plus a free preview of the cards already on file. Measurement, not certification — never a rank, a grade or a certificate. Verification stays free.
Inputs: subject (required), axis, x_payment.
get_axisAInspect
One axis row from the live GSPC board at https://councilof.ai/api/gspc: n, accuracy, interval, MEASURED or UNMEASURED status, family, and dates. An unmeasured axis is a first-class answer — a declared slot with no run behind it, published so the gap is visible — never an error and never a zero. Never a certification.
Inputs: axis (required).
get_cardAInspect
GET one public-root card-v0 leaf by sha256 (64 hex) from https://councilof.ai/cards/{sha16}.json. UNMEASURED cells stay visible. Three states: VALID (fetched), INVALID (not a leaf of the live root), UNCHECKABLE (fetch failed). Not a GSPC measurement-card.
Inputs: sha256 (required).
get_rootAInspect
GET the permissionless public-root at https://councilof.ai/root.json. Returns merkle_root, card_count, as_of, and UNMEASURED notes. Separate from GSPC. Three states: VALID (fetched), UNREACHABLE (could not fetch), UNCHECKABLE (body unreadable). Never a certificate.
Inputs: no named parameters.
list_cardsAInspect
The published signed-card index (https://councilof.ai/signed/card_index.json): what the index declares (n_cards) and how many rows it actually carries, reported next to — never reconciled with — the count the card store endpoint (https://councilof.ai/api/cards) reports for itself. Two labelled numbers from two surfaces; if they disagree, this tool shows the disagreement rather than picking one. Optional filters return recent rows: axis, limit.
Inputs: axis, limit.
receipts_batchAInspect
PAID (x402 or CSOAI LTD invoice). A historical batch of the estate's measurement receipts via https://councilof.ai/api/receipts/batch: every signed card-v0 leaf whose as_of falls in [from,to] (≤200), each with its Merkle inclusion path and the public root(s) that carried it, plus the root index for the window and one signed manifest card citing the batch sha256. preview=true is free and returns count, span, root count and the sha256 of the exact bytes the paid call returns. Recent leaves are free at /root.json, /cards/ and /api/proof?sha= — the batch sells assembly across history, never a conclusion. No settlement-receipt stream exists (/api/receipts/latest is UNPUBLISHED) and this tool never claims one. Without x_payment the tool returns the 402 challenge.
Inputs: from (required), to, preview, x_payment.
rwa_evidenceAInspect
PAID (x402). Per-request signed evidence card of ONE XRPL issued asset's deterministic on-ledger state via https://councilof.ai/api/rwa/evidence: AccountRoot lsf* flags, Domain, the two-way xrp-ledger.toml check (PASS / FAIL / UNCHECKABLE — unreachable is never FAIL), gateway_balances obligation, holders as the free reader has them, every raw fetch sha256'd. preview=true is free (unsigned state). Historical state at fetched_at — not a rating, not a guarantee, not a conformity mark; /api/xrpl and /root.json stay free. Without x_payment the tool returns the 402 challenge. If the route is not deployed on this origin the tool says NOT_DEPLOYED.
Inputs: asset (required), preview, x_payment.
verify_cardAInspect
Verify a signed gspc.measurement-card under the published rule (https://councilof.ai/signed/HOW-TO-VERIFY.md): recompute the id from the canonical body bytes, then check the Ed25519 signature under the PINNED key published at did:web:csoai.org#card-attestation-1. A card that carries its own key proves only that the file is self-consistent — anyone can alter a body and sign it with a key they just generated — so a signer other than the published key is reported INVALID. Three verdicts, never two: VALID, INVALID (with the reason), or UNCHECKABLE when the check could not be completed — 'could not check' is a different claim from 'forged'. Accepts the card as a JSON object, a JSON string, or a councilof.ai / csoai.org URL. Never a certification.
Inputs: card (required).
verify_inclusionAInspect
Check a sha256 against the live public-root merkle via GET /api/proof?sha=. Three states only: VALID (included), INVALID (not a leaf), UNCHECKABLE (proof endpoint unreachable). Does not claim Ed25519 unless sig_ed25519 is present and checked separately. Never a grade.
Inputs: sha256 (required).
x402_trustAInspect
GET the latest x402 catalog trust snapshot: counts of how many catalogued x402 resources open a correct 402 challenge vs how many are phantom on the wire. Counts only by doctrine — host details withheld; a 402 is NOT delivery; measurement, never certification.
Inputs: no named parameters. Returns: as_of, counts, headline
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 12 tools
Most tools target distinct resources/actions (root fetch, card fetch, inclusion proof, card signature verification, board totals, single axis, index listing), and descriptions explicitly distinguish states and boundaries. However the term 'card' is overloaded across card-v0 leaves, signed measurement-cards, and commissioned receipts, so get_card/verify_card/list_cards/commission_card can blur if read quickly. The paid evidence tools (art50, rwa, receipts_batch, commission_card) are differentiated by domain but share the x402 payment framing.
Roughly half the tools follow a clean verb_noun snake_case pattern (get_root, get_card, verify_inclusion, get_axis, list_cards, verify_card), but the rest are bare noun phrases (x402_trust, art50_marking_evidence, rwa_evidence, receipts_batch, board_totals). All snake_case and readable, yet two mixed conventions coexist. No truly chaotic naming, just an inconsistent verb-vs-noun approach.
12 tools sits comfortably in the well-scoped 3-15 band. The count is slightly pushed by the server bundling several sub-services (merkle/root infra, GSPC board, x402 trust, Art 50, XRPL RWA, receipts), so it feels broad rather than tight, but each tool maps to a real operation. No padding or extreme mismatch.
Core lifecycle is covered: read root, read a leaf, verify inclusion, list the index, verify signed cards, get board totals and a single axis, plus paid attestation/evidence generation with free previews and explicit NOT_DEPLOYED/UNCHECKABLE fallbacks. Minor gaps exist (no 'list_axes' to enumerate all axis rows, x402_trust yields counts only, no subject search), but these are workable around. Surface is substantially complete for a measurement/attestation service.
Maintenance
Related MCP Connectors
Register every AI agent, log every action, prove it. EU AI Act compliance built in.
Runtime AI governance: decision gates, human approval, hash-chained audit, compliance mapping.
AI inventory and EU AI Act compliance. Find AI tools and use cases, check new AI uses before launch.
AI governance MCP server for EU AI Act compliance and jurisdiction verification
Related MCP Servers
- AlicenseAqualityDmaintenanceProvides automated EU AI Act compliance tools, including risk classification, role determination, transparency disclosures, content watermarking, deepfake labeling, and security threat detection.1635Apache 2.0
- AlicenseBqualityCmaintenanceManages trust chains and attestations with built-in EU AI Act compliance.55 npmMIT
- AlicenseNot gradedqualityDmaintenanceEnables EU AI Act compliance for AI agent systems by providing risk classification, audit trails, gap analysis, and evidence package generation.55 npmMIT
- AlicenseNot gradedqualityCmaintenanceProvides AI content watermarking and C2PA compliance for EU AI Act Article 50, enabling detection, verification, and batch processing of authenticated content.7 npm29 PyPIMIT