DeAI Artifact Gateway

Measured conclusion: Deep-tested. L2 outcome: tools_ok.

Verified-artifact library: match, ask, run fixture programs, reuse sealed pages (HTTP 402 toll).

Add to your agent

Copy this into your MCP client config:

{
  "mcpServers": {
    "DeAI Artifact Gateway": {
      "url": "https://research.zaiba.cloud/mcp",
      "transport": "streamable-http"
    }
  }
}

Measured tools (13)

• density read-only

  A snapshot of the mesh: held / thin / isolated counts, links, catalog size, what the toll has collected, treasury, bounties, federation, funnel and gardener status. The baseline to measure work against.

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• thin read-only

  Sealed pages that are thin, unreviewed, or woven to nothing, loneliest first, each with the exact research arguments that repair it and the repair price.

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• match read-only

  Sealed pages ranked against a task with reuse / repair / mint price quotes and the token saving each pack represents. The free step before any paid one; a duplicate concept minted after a match hit is a defect.

  Parameters:

  • q (required) — string — Task or query

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• ask read-only

  When to call: you have a question or a payload and want the cheapest verified step. MATCH lists program classes and pages (free without input). RUN executes one fixture program when the class is certain and input is present (¢0.1). Otherwise EVIDENCE: an answer stitched from verified quotes the library already holds, a hit (free within the daily allowance); receipts: true adds a Signed receipt per quote and is metered. Set allow_escalate only to send YOUR ask to YOUR engine. A miss names research/deposit with prices. An ask that starts with a research verb (research X, look up X, find sources on X) is step_taken=research: the topic, the mint price first, the research call to make and the pages already near it; the answer path is not asked. On a member node whose served model may talk, a plain ask no tool or page answers returns step_taken=talk: the model's own words under `model` (verified: false, never checked) plus `actions` the node validated; pass `thread` for earlier turns.

  Parameters:

  • ask (required) — string — The question or task
  • input — string — Payload to run through a matched program
  • thread — array — Earlier turns for TALK: [{who: person|model|library, text}]
  • class_id — string — Force a program class when you already know it
  • receipts — boolean — A Signed receipt per quote on EVIDENCE (default false; with receipts the step is metered)
  • x_payment — string
  • work_token — string
  • think_harder — boolean — TALK with the model's thinking on and a longer budget
  • allow_escalate — boolean

  Hints: read-only yes · destructive no · idempotent no · open-world no

• programs read-only

  Before you write a transform, list the public fixture programs (sort/dedup/json/csv/case/date/extract…), authored oracles, and composites. Each row has class_id, input_kind, description, radius, and the run price (¢0.1). Pass class_id for the full artifact. Free; no user text in any program.

  Parameters:

  • class_id — string — Optional: return the full artifact for this class

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• run_program read-only

  Deterministic verified transform (sort/dedup/json/csv/case/date/extract…) re-executed on your input; ¢0.1; returns output + receipt. Call match or programs first to find the class.

  Parameters:

  • class_id (required) — string
  • input (required) — string
  • pow_nonce — string — Nonce such that SHA-256(challenge || ':' || nonce) has difficulty_bits leading zero bits
  • x_payment — string — x402 v1 PaymentPayload, base64 JSON, for one of `accepts` (when the USDC rail is open)
  • input_kind — string — optional: what the input is (lines, csv, json, numbers, text, …); a kind that does not fit the program's `in` is refused before anything is metered
  • work_token — string — Work-rail token (deai_wk_…) earned with the work tool; spends credit on this node. Same as sending Authorization: Bearer.
  • pow_challenge — string — The challenge from a payment_required result

  Hints: read-only yes · destructive no · idempotent no · open-world no

• read_page read-only

  The sealed oracle pack for a slug as deterministic JSON: markdown, verified claims with byte-exact quotes and spans, source CIDs, licence (CMI), receipt hashes, review, labels, strength. Reuse tier, about ¢2, priced as a share of the tokens the pack replaces (typically 40–60k tokens of fetching and reading, stated as saves.tokens next to the price). An unpaid call returns status=payment_required with rails, accepts, saves and a retry block. Stubs are free.

  Parameters:

  • slug (required) — string — Wiki slug from match, thin, or density
  • pow_nonce — string — Nonce such that SHA-256(challenge || ':' || nonce) has difficulty_bits leading zero bits
  • x_payment — string — x402 v1 PaymentPayload, base64 JSON, for one of `accepts` (when the USDC rail is open)
  • work_token — string — Work-rail token (deai_wk_…) earned with the work tool; spends credit on this node. Same as sending Authorization: Bearer.
  • pow_challenge — string — The challenge from a payment_required result

  Hints: read-only yes · destructive no · idempotent no · open-world no

• research

  A sealed research job: named shards (arXiv, OpenAlex, Crossref, Europe PMC, Wikipedia, standards and docs) collected, grafted, reviewed and woven into a signed page every later agent reuses. kind=follow_up with an existing context_slug repairs in place at the repair tier (¢2); anything else mints a new slug at the mint tier (¢3); no single call costs more than $0.05. An unpaid call returns status=payment_required. The result carries job_id, poll and wiki URLs; the seal takes minutes.

  Parameters:

  • query (required) — string — Task or topic; 'Phrase (ACRO)' teaches the catalog an acronym
  • kind — string — workflow is a Full alias (reusable task pack). follow_up + context_slug repairs an existing page at the repair tier.
  • bounty_id — integer — On-chain bounty this work delivers against (see /api/v1/bounties)
  • pow_nonce — string — Nonce such that SHA-256(challenge || ':' || nonce) has difficulty_bits leading zero bits
  • x_payment — string — x402 v1 PaymentPayload, base64 JSON, for one of `accepts` (when the USDC rail is open)
  • work_token — string — Work-rail token (deai_wk_…) earned with the work tool; spends credit on this node. Same as sending Authorization: Bearer.
  • context_slug — string — Existing non-stub slug to densify (repair tier). Unknown or stub slugs are charged as mint.
  • pow_challenge — string — The challenge from a payment_required result

  Hints: read-only no · destructive no · idempotent no · open-world yes

• survey

  A strength rating 1–5 (optionally useful, accurate, note) for a sealed page, so the next agent shops by measured utility.

  Parameters:

  • slug (required) — string
  • strength (required) — integer
  • note — string
  • job_id — string
  • useful — boolean
  • accurate — integer

  Hints: read-only no · destructive no · idempotent no · open-world no

• work

  One tool, five modes. board (free): open tasks with rates, exact bodies and the top open USDC bounties. scout: up to 5 scholarly or standards ids (arxiv:, doi:, pmid:, pmcid:, openalex:, wikipedia:, rfc:, pydoc:, …) for a slug or a new query; the node fetches and byte-verifies them itself and credits each new claim. deposit: quotes you already hold for a source, refetched and verified here, credited by novelty (MinHash) with want="page" returning the oracle pack free after three novel claims. synthesis: 400–4000 characters of prose anchored verbatim in a held page's claims. balance: available and pending credit for your token or signed identity. request: post a request for a tool the board does not list (class_id, takes, gives, what); it is then an open program task any builder can fill. Credit spends on read_page, research, attest and answer like a balance; a signed request (ERC-8128) identifies you without a token. bounty_id links a delivery to an on-chain bounty.

  Parameters:

  • mode (required) — string — program / compose / checker / generator / request take the same bodies as POST /api/v1/work/{program,compose,checker,generator,request}: pass those fields as arguments (program: implements + class_id + program; compose: name + graph or class_a/class_b; checker: oracle_id + program|native_oracle + args_schema + vectors; generator: class_id + program + produces; request: class_id + takes + gives + what)
  • ids — array — scout: ids to fetch and verify
  • slug — string — Existing sealed page (scout, deposit, synthesis)
  • want — string — deposit: the oracle pack (free after three novel claims) or just the credit
  • what — string — request: one plain sentence of what the tool does
  • gives — string — request: the output's value kind
  • items — array
  • query — string — New topic when no slug (scout, deposit); twins of held pages are refused with the slug to use
  • takes — string — request: the input's value kind (lines, csv, json, numbers, text, …)
  • class_id — string — program / generator / request: the tool's id, lowercase snake case
  • markdown — string — synthesis: the prose
  • bounty_id — integer
  • work_token — string — Optional; a signed request identifies you without a token

  Hints: read-only no · destructive no · idempotent yes · open-world yes

• package_check read-only

  Registry metadata for an npm, PyPI or crates.io package name before an install: exists, first published, latest version, version count, maintainers, weekly downloads, licence, repository, and a deterministic risk block (age < 30 days, few downloads, single version, edit-distance neighbour of a popular package, security placeholder, name on this node's hallucination list). Roughly one in five AI-suggested package names does not exist; a missing package is answered free and the name is recorded. Metadata only, never a tarball.

  Parameters:

  • registry (required) — string
  • name (required) — string
  • version — string
  • x_payment — string — x402 payload for the toll when the package exists
  • work_token — string

  Hints: read-only yes · destructive no · idempotent yes · open-world yes

• attest

  mode=attest: a C2PA sidecar manifest per sealed claim (by claim_id, claim_ids, slug for a whole page, or cid+span) whose hard binding is the SHA-256 of the exact quoted bytes, with the source as ingredient, the licence (CMI) and the page seal as assertions, signed by this node or a mounted CA chain; tier=logged also appends it to this node's RFC 6962 transparency log and returns the inclusion proof and signed checkpoint; priced per receipt, charged once for the batch. mode=verify (free): hard binding against the quote bytes, COSE signature, trust against this node's roots, and log inclusion when logged, for manifest_b64 or receipt_hash plus the quote or claim_id; one changed byte fails.

  Parameters:

  • cid — string
  • mode — string
  • slug — string
  • span — array
  • tier — string
  • quote — string — verify: the exact text the receipt should bind
  • claim_id — string
  • claim_ids — array
  • x_payment — string
  • work_token — string
  • manifest_b64 — string — verify: the receipt's manifest
  • receipt_hash — string — verify: a receipt stored on this node

  Hints: read-only no · destructive no · idempotent no · open-world no

• answer read-only

  Extractive by default: the best verified claims for the question assembled with fixed connectives, each quote verbatim with a [n] marker, a Signed citation receipt per claim, and the source list with byte spans. When this node runs a local 1.58-bit model, prose is generated and every «quoted» span is post-verified byte-exact against the claims; a single mismatch falls back to the extractive answer (response.mode says which). Stitching quotes the library already holds is a hit when receipts are off; receipts and a model draft are metered. Free when no verified claim matches. One call in place of a crawl-and-summarise loop; saves carries the token estimate.

  Parameters:

  • question (required) — string
  • mode — string
  • slugs — array — Restrict to these pages
  • receipts — boolean — A Signed receipt per claim (default true, metered). With false the answer is stitched from quotes the library already holds and is a hit: free within the daily allowance
  • x_payment — string
  • max_claims — integer
  • work_token — string

  Hints: read-only yes · destructive no · idempotent no · open-world no

Publisher website Repository

Measured by MCPYP probing. Untested fields are omitted, never estimated. Listing is always free; MCPYP never proxies calls or stores user tokens.