Prototype · Preview: every server in the official registry, synced every 30 minutes. Sign in with Google to write reviews.
PopNx MCPMCP servers for people who use AI · explained in plain words

Keryx

by tang-vuversion 0.4.48i★ 3iComputer or cloud
Registry nameio.github.tang-vu/keryx

How to install

Overview

Description from the maker (original English)

Budgeted research with cited evidence and visible payment state on the selected Arc network.

What you can do with it

We have not written a plain-language summary for this one yet. Below are the facts we read from the registry.

Install

Pick the AI app you use, then copy the settings.

Passed 4/5 checks · What you ask goes through the maker’s server (keryx.cc)

First: which AI do you use?

Pick your app and the steps appear below.

Works with Claude · about 2 min

No key, nothing to install · Set it up once on claude.ai or the desktop app, then use it in chat · free accounts can add 1

  1. Open Claude’s Connectors page

    CustomizeConnectors

    CustomizeConnectors1+ Add2https://…3Add4
  2. Click “+ Add” → “Add custom connector” and fill in these 2 fields

    Copy each value into the field with the same name in Claude, then click Continue

    Name · NameKeryx
    Remote MCP server URL · Addresshttps://keryx.cc/mcp

    Leave Authentication as Claude detected it, then click Add

  3. Turn it on in a chat

    +ConnectorsKeryx

  4. Try it

    Use Keryx and tell me what you can do for me, then show me one thing

    ✓ If Claude says it is using Keryx or shows a Keryx tool in the answer, you are connected

Menu names checked against Anthropic’s help page on Nov 11, 2026 · your screen may differ slightly by app version

Works with ChatGPT on business plans only

ChatGPT Business, Enterprise or Edu on the web, added by a workspace admin or owner · On Free, Plus or Pro: your ChatGPT plan cannot connect MCP yet

  1. Open ChatGPT’s add-MCP page on the web

    Plugins+Add custom MCP server

    ChatGPTPlugins1+2https://…3Create4
  2. Fill in the name and address

    Enter a name, paste the address under Connection, tick the risk warning and click Create as a plugin

    Name · NameKeryx
    Connection · Addresshttps://keryx.cc/mcp
  3. Pick Keryx in a chat

    Start a new chat and pick Keryx from your plugins before typing

  4. Try it

    Use Keryx and tell me what you can do for me, then show me one thing

    ✓ If ChatGPT says it is using Keryx or shows a Keryx tool in the answer, you are connected

Menu names checked against OpenAI’s help page on Nov 11, 2026 · your screen may differ slightly by app version

Works with Claude Code · about 2 min

No key, nothing to install

  1. Open Terminal

    Mac: press Command+Space and type Terminal · Windows: open PowerShell

  2. Paste this command and press Enter

    claude mcp add --transport http keryx https://keryx.cc/mcp
  3. Type /mcp in Claude Code

    You should see Keryx listed as connected

  4. Try it

    Use Keryx and tell me what you can do for me, then show me one thing

    ✓ If Claude Code says it is using Keryx or shows a Keryx tool in the answer, you are connected

Menu names checked against Anthropic’s help page on Nov 11, 2026 · your screen may differ slightly by app version

Works with Cursor · about 2 min

No key, nothing to install

  1. Click this button to open Cursor

    Open in Cursor

    Cursor asks you to confirm; click Install

  2. Open the AI chat in Cursor in Agent mode

  3. Try it

    Use Keryx and tell me what you can do for me, then show me one thing

    ✓ If Cursor says it is using Keryx or shows a Keryx tool in the answer, you are connected

Menu names checked against Cursor’s help page on Nov 11, 2026 · your screen may differ slightly by app version

Works with VS Code · about 2 min

No key, nothing to install

  1. Click this button to open VS Code

    Open in VS Code

    VS Code asks you to confirm; click Install

  2. Open the AI chat in VS Code in Agent mode

  3. Try it

    Use Keryx and tell me what you can do for me, then show me one thing

    ✓ If VS Code says it is using Keryx or shows a Keryx tool in the answer, you are connected

Menu names checked against VS Code’s help page on Nov 11, 2026 · your screen may differ slightly by app version

Other apps that use MCP, such as Windsurf, Zed or Gemini CLI, each add servers differently. See how in the app list.

See every app that uses MCP →

Address for apps that connect over the web

Addresshttps://keryx.cc/mcp

JSON settings are under “For developers” below

Stuck? Ask people who use Keryx💬 Discussion →
For developers: Terminal commands and JSON settings

Over the web

Claude Code

claude mcp add --transport http keryx https://keryx.cc/mcp

Cursor (~/.cursor/mcp.json)

{
  "mcpServers": {
    "keryx": {
      "url": "https://keryx.cc/mcp"
    }
  }
}

VS Code (mcp.json)

{
  "servers": {
    "keryx": {
      "type": "http",
      "url": "https://keryx.cc/mcp"
    }
  }
}

Installed on your computer

Claude Code

claude mcp add keryx -- npx -y keryx-mcp

Claude Desktop (claude_desktop_config.json)

{
  "mcpServers": {
    "keryx": {
      "command": "npx",
      "args": [
        "-y",
        "keryx-mcp"
      ]
    }
  }
}

Cursor (~/.cursor/mcp.json)

{
  "mcpServers": {
    "keryx": {
      "command": "npx",
      "args": [
        "-y",
        "keryx-mcp"
      ]
    }
  }
}

VS Code (mcp.json)

{
  "servers": {
    "keryx": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "keryx-mcp"
      ]
    }
  }
}

These settings are generated from the maker’s registry entry. If the README says otherwise, follow the README.

At a glance

Downloads per weeki
1,097/wk
Licencei
No licence stated
Connection
Computer or cloud
Transport
stdio, streamable-http
Latest versioni
0.4.48
Published to registryi
Nov 10, 2026
Code last changedi
Nov 10, 2026
Maker
tang-vu
Packages
npm: keryx-mcp
Website
keryx.cc
Registry status
Active

Which apps can use it

Based on how this server connects and on each app’s own docs, checked Nov 10, 2026.

Connect over the internet in

ClaudeChatGPTGooseLM StudioMsty StudioCherry StudioCursorVS Code + GitHub CopilotVisual StudioWindsurf (Devin Desktop)ZedKiroClineContinueJetBrains AI AssistantClaude CodeCodex CLIGemini CLIOpenCodenodecodeWarpLibreChatOpen WebUIRaycast

Run on your computer in

ClaudeGooseLM StudioMsty StudioCherry StudioCursorVS Code + GitHub CopilotVisual StudioWindsurf (Devin Desktop)ZedKiroClineContinueJetBrains AI AssistantClaude CodeCodex CLIGemini CLIOpenCodenodecodeWarpLibreChatRaycast

All apps, with pros and cons →

Related guides

What it can do

Tool names and descriptions exactly as the server reported them on Nov 11, 2026. The maker wrote them, not us; we show them as plain text.

  • evidence_draft

    Assemble user-selected claim spans and related-work quote themes from saved Include paper records and caller-supplied retained Keryx report rows. A theme may carry author-written statements (contribution, agreement, disagreement, gap); a statement naming a paper is exported only

  • research_feedback_read

    Original-caller private opinion on the exact delivered ordinary research answer. Requires its private bearer capability; a public report ID or link grants no authority. First choice is retained, exact retries are idempotent. Silence is no response. No research, source fetching, r

  • research_feedback_submit

    Original-caller private opinion on the exact delivered ordinary research answer. Requires its private bearer capability; a public report ID or link grants no authority. First choice is retained, exact retries are idempotent. Silence is no response. No research, source fetching, r

  • draft_check

    Split a private passage into sentences, bind citation markers ([3], cite{key}, [@key], Author (Year)) to a BibTeX, RIS or DOI/link reference list, and compare each cited sentence with source text you supply. Each verdict (supported, partly-supported, not-found, source-unavailable

  • evidence_procurement

    Assess missing evidence requirements, supplied provider offers, selections, deliveries and caller reviews with the shared private evidence procurement contract. Amounts are exact integer micro-USDC. Signature and content-binding checks do not establish delivery quality, buyer aut

  • deliverable_acceptance_read

    Owner-only ordinary prepaid A2A deliverable state. Requires explicit deliverable:read. Historical remedy:none and missing response window remain visible. Native/enrolled, Monthly, private v2 and browser-only originals are unavailable. Not source-decision agreement or semantic ver

  • deliverable_acceptance_submit

    Record Accept, Request revision or Reject for the original customer's exact retained delivery. Requires explicit deliverable:write, expected current revision/digests and an idempotency key. Reason is private; publishState explicitly shares state/timestamp only. Request storage do

  • operator_obligations_read

    Read a private conservative obligation projection for the exact server-delegated reader with explicit operator:read API-key scope. Disabled without protected reader/role configuration. Partial native history remains unknown with zero safe/advisory surplus. No cash probe, signer,

  • paper_lookup

    Find paper titles, recorded contributors, DOI and exact repository versions without research or payment. Default: retained catalog metadata with observation times. Set searchRepositories=true only to explicitly send the query to arXiv/Crossref (at most two requests, no paper-body

  • research

    Research a question under a USDC creator-payment budget. Keryx selects sources, pays access tolls and weighted citation rewards on the configured Arc network, then returns qualified source excerpts and a receipt. Complete synthesis and per-assertion entailment remain unverified.

  • keryx_status

    Describe this MCP surface, its payment behavior, and the active caller tier.

  • research_monthly

    Read the four-request, 30-day Research Monthly quote and handoff. The current quote identifies its payment network. Never buys or redeems a plan.

  • keryx_operator_status

    Observe public prepaid research operations, queue, recorded ordinary/recovered/unknown completion latency cohorts and hold/review rationale. Timing is order createdAt to updatedAt (acceptance to recorded completion/update); later reconciliation can extend it. It is not first-answ

  • keryx_public_job_ledger

    Inspect a partial selected-network public web-dispatch transfer ledger and balanced exact micro-USDC vouchers. Browser funding and Keryx sponsorship are separate. Pending/ambiguous/batched-attribution legs do not enter settled totals. No customer questions/identities, private inv

  • keryx_purchase_outcomes

    Read a public dispatch's retained exact-version BUY/citation/payment observations. Returns bounded sample counts, citation hit rate, calibration bands and integer micro-USDC uncited access cost. Recorded settlement is not revalidated on chain. Historical participant cohort is unk

  • keryx_source_learning

    Read the same bounded observational source/topic record as the public creator page and API. Only write-time independently classified/funded outside observations qualify; owners, unknown and private history are excluded. Source-declared public topics, counts, recorded citation con

  • profile_read

    Read the verified key owner's private profile and attributed activity. Requires explicit profile:read scope. No public identity, research or payment.

  • profile_update

    Replace the verified key owner's private display name, handle, one-line bio/purpose and links. Requires explicit profile:write scope. No wallet selector or identity verification; no payment/history changes.

  • history_read

    Read bounded pages of the verified API-key wallet's ordinary current-store runs. Requires explicit history:read. Literal case-sensitive question search; storeNetwork describes the deployment, not per-run settlement. Recorded spend/allocation is not a payment receipt. No wallet se

  • history_owned_read

    Same owner-private v2 contract as /api/me/asks. Explicit history:read required. Current store, bounded filters/keysets; missing membership/claims stay explicit. Export has explicit truncation/continuation. No research, signer, payment, public-owner selector or mutation authority.

  • history_export

    Same owner-private v2 contract as /api/me/asks. Explicit history:read required. Current store, bounded filters/keysets; missing membership/claims stay explicit. Export has explicit truncation/continuation. No research, signer, payment, public-owner selector or mutation authority.

  • history_private_jobs_list

    Private data for this authenticated caller only. Exact-payer completed private jobs in the current store; explicit history:read required. Bounded scan/continuation, unknown provenance and NULL own spend/creator settlement remain explicit. No wallet selector, public report/share l

  • history_private_jobs_detail

    Private data for this authenticated caller only. Exact-payer completed private jobs in the current store; explicit history:read required. Bounded scan/continuation, unknown provenance and NULL own spend/creator settlement remain explicit. No wallet selector, public report/share l

  • history_private_jobs_export

    Private data for this authenticated caller only. Exact-payer completed private jobs in the current store; explicit history:read required. Bounded scan/continuation, unknown provenance and NULL own spend/creator settlement remain explicit. No wallet selector, public report/share l

  • run_privacy_read

    Owner-private current-store read using the verified key's explicit history:read scope. No wallet selector or signing. Visibility, defaults and erasure require the interactive web session. Storage admission is required.

  • run_owner_export

    Owner-private current-store read using the verified key's explicit history:read scope. No wallet selector or signing. Visibility, defaults and erasure require the interactive web session. Storage admission is required.

  • profile_identities_read

    Read only the verified key owner's saved ORCID/GitHub identities. Requires explicit profile:read scope. No wallet selector, provider OAuth, linking, unlinking, research or payment. Link identities interactively in your private profile browser page.

Basic checksi

We check information anyone can see; we have not read the code. Checks that do not apply are left out of the count. · Checked Nov 12, 2026

  • ✓ Pass

    Maker matches the repo

    The registry name was verified with the GitHub account tang-vu, which owns the repo

  • ✓ Pass

    Still maintained

    Code last changed yesterday

  • ✕ Fail

    Has a licence

    No licence file found in the repo

  • ✓ Pass

    Endpoint responds

    Responded in 557 ms (HTTP 200)

  • – Not applicable

    Sign-in with OAuth

    No sign-in needed

  • ✓ Pass

    No hidden instructions in tool descriptions

    Read 27 tool descriptions and found no hidden instructions to the AI

Discussion

Ask, share a tip or report that it does not work. Sorted by what people marked helpful.

No posts yet. Be the first to ask or share.

Posts here are written by users, not the PopNx team.

Sign in with Google to post

User reviews

No reviews yet

Sign in with Google to write a review

Compared with others in Online shops & paymentsi

  • More stars than 95% of the other 2721 servers in this category
  • Published to the registry more recently than 91% of the other 2721 servers in this category
  • Basic checks equal to or better than 62% of the other 2721 servers in this category

Compared with every active server in this category in the registry. Stars count only when the publisher owns the repo.

Common questions

Is Keryx safe?

It passed 4 of 5 applicable basic checks. We have not read the code. Before adding a key or token, see what access the server asks for and use a key with only the permissions you need.

What keys or codes does it need?

The registry does not list a required key, though some servers ask you to sign in on first use.

Does it run on my computer or in the cloud?

Either way works: install it on your computer or connect to the cloud version.

Is it still maintained?

The GitHub repo last changed yesterday, and the latest version (0.4.48) reached the registry yesterday.

Which apps can use it?

Apps that support MCP, such as Claude Desktop, Claude Code, Cursor and VS Code. It connects by URL, so apps that accept remote MCP servers work too, such as ChatGPT Business and above.

What can I use instead?

In the same category: Spree Commerce developer docs for AI agents, Stripe MCP, Sessy — Amazon SES observability

Alternatives

Servers in the same category from other makers, sorted by stars.

Reading a card:↓ Downloads per weeki★ GitHub starsiThree pillars: reviews · checks · usageiUpdated = latest version published to the registryi

Spree Commerce developer docs for AI agents

Spree Commerce docs for AI coding agents: build headless B2B and marketplace ecommerce on its APIs.

by spreeWidely used★ 15.8K Stars for the whole repoUpdated 12 days ago

Runs in the cloud

Stripe MCP

Stripe’s official server lets AI view and manage customers, products, payments and invoices. Connect to Stripe’s cloud and sign in with OAuth.

by stripeWidely used★ 2.1KUpdated 13 months ago

Runs in the cloud✓ stripe.com

Sessy — Amazon SES observability

Read-only Amazon SES observability: search events, inspect bounces, pull delivery stats.

by marckohlbruggeGrowing★ 945Updated 3 months ago

Runs in the cloud

README from the maker

Open the README

Copied from the maker’s repo as written. PopNx MCP has not reviewed or endorsed it. HTML, images and scripts are removed. Links lead off this site.

Keryx MCP

Inspect a private saved library locally (source candidate)

inspect_saved_literature({backup, baselineBackup?}) validates a caller-supplied personal-bibliography v1 backup and returns aggregate counts for saved records, personal screening, repositories, recorded publication kinds and missing metadata. Output omits private focus, notes and individual paper identities. Missing fields and current paper status remain unknown; counts are not reading, quality or settlement proof. This local tool has no file, network, model, wallet or payment I/O. MCP clients may retain the entire supplied backup, including private notes. See offline inspection usage and limits.

Optional baselineBackup adds aggregate changes from that before snapshot to the backup after snapshot. Exact URLs define retained records; versions are not inferred matches. Metadata, timestamps and personal-state changes stay separate, and no changed values or paper identities enter the result. Both snapshots must validate within their individual 1 MiB limits. Clients may retain both full inputs. Omitting baselineBackup keeps the existing single-backup response.

Export private saved literature locally (source candidate)

export_saved_literature({backup, format, screening?, paperUrls?}) converts a caller-selected personal-bibliography v1 backup to BibTeX, RIS or CSL-JSON. It validates the full 1 MiB bounded JSON before selecting saved screening states, keeps exact versions and provenance, and omits private review title, question, notes and decisions. Optional paperUrls selects at most 50 unique exact saved URLs in caller order; an empty array exports no references. Missing/changed-version URLs refuse the entire export, and screening filters within the selection. No URL is fetched. This local tool has no file, network, model or payment I/O. Your MCP client may retain the entire supplied backup, including private notes. See offline export usage and boundaries.

The CLI bundles the pinned Circle 3.5.0 buyer implementation; its exact inputs are recorded in dist/build-inputs.json. Its application validators use the bundled, security-patched Zod 3 compatibility runtime, retaining existing UTF-16 limits and schema behavior. MCP SDK, its published Zod 4 dependency and viem remain external. The buyer does not auto-install x402's old Zod peer graph. Source builds require the repository's pinned dependencies. Bundled code, modifications and licenses are listed in THIRD-PARTY-NOTICES.txt.

Compare saved receipts locally (source candidate)

compare_saved_research({before, after, matches?, matchingPlan?}) accepts two version-1 research receipt JSON strings, at most 512 KiB each. It verifies both recorded payload checksums and compares exact retained excerpt rows, source versions and reading context. Unique identical targets are paired automatically; explicit one-to-one matches use recorded claim indices. [] leaves all targets unmatched. Alternatively, matchingPlan takes saved keryx-report-matching-v1 JSON (up to 64 KiB), bound to those exact receipt payload checksums in that order. Do not combine it with matches. The result includes a reusable matchingPlan object containing only checksums and target indices, alongside comparison. Missing historical ledgers stay unknown. A checksum is not authenticity, factual support, settlement or proof of improvement; row counts are not independent sources.

This stdio tool uses no network, model, database or payment capabilities. Your MCP client can retain the supplied text and results. For browser-memory-only file inspection and JSON/Markdown/CSV downloads, use /research/compare. API/remote MCP, extension and bot receipt downloads hand off to that local workspace. See the comparison contract for local CLI, matching, limits and export semantics. Production/package availability follows the coordinated release; this source section does not claim an installed version.

Sponsored public research (source candidate)

ask_keryx_sponsored runs hosted sponsored research with no API key, buyer wallet or funding setup. First build the source checkout:

npm install

npm --prefix mcp run build

Configure your MCP client to start the resulting bundle using its absolute path:

{

"mcpServers": { "keryx": { "command": "node", "args": ["/absolute/path/to/keryx/mcp/dist/keryx-mcp.mjs"] } } }

This configuration also works when the MCP client starts from another directory. Explicitly consent to a public question and trial report when calling the tool:

ask_keryx_sponsored({

question: "What evidence supports sodium-ion batteries for grid storage?", publicReport: true, mode: "quick", budget: 0 })

Questions are bounded to 2,000 characters; mode is quick or deep. The optional source budget accepts 0–0.08 USDC, including zero, independently of model/search operating costs. Server quotas, caps and pauses apply. No private task fields are accepted. A timeout or disconnect may leave server work running: there are no automatic retries or automatic paid fallbacks. Quota refusals require waiting before a deliberate new submission. Returned payment evidence keeps planned, pending and settled states distinct.

The separate ask_keryx tool remains the paid buyer path and retains its own wallet, merchant policy and recovery requirements. This addition is a source candidate; package publication and installed availability belong to the parent release and are not claimed here.

The ORIGINAL268 NEXT source candidate adds history_private_jobs_list, history_private_jobs_detail and history_private_jobs_export to both MCP registrations. Current published 0.4.20 does not contain these tools. No new package/version, artifact URL, installed availability or production delivery is claimed. Parent review, focused checks, required CI and exact-source packed MCP acceptance/publication remain required.

In this source checkout, the stdio registration uses an explicit KERYX_API_KEY whose live server key has history:read. It calls only the existing HTTPS POST /api/me/private-jobs, /detail and /export JSON contracts. No wallet private key, buyer wallet, cookie, chat/client label or expected-wallet precondition grants history ownership. The expected wallet can only refuse a mismatch. The remote registration uses the verified request actor and original request-local history revalidator before Db acquisition, after it and after private reading.

These are owner-private responses; an MCP desktop/agent client may retain tool messages. List is a finite exact-payer completed-job scan; detail is an explicit private answer read without public report/share links or automatic research/model submission; export returns bounded JSON metadata with complete, truncated and continuation. Own settled spend, creator rewards and paid creators remain NULL, and missing recorded provenance stays unknown. No research, source fetch, signer, purchase, database mutation, retry or alternate credential is used by these tools. Existing paid tools retain their separate roles.

The source-only CLI adapter is available at scripts/private-job-history.mts:

node --import tsx scripts/private-job-history.mts --help

node --import tsx scripts/private-job-history.mts list --limit 25 node --import tsx scripts/private-job-history.mts detail --id 'prv_<64-lowercase-hex>' node --import tsx scripts/private-job-history.mts export --format csv

The last three commands print private data for the authenticated caller. Supply the history key through the caller's existing private environment; never paste it into flags, URLs or a model prompt. The CLI loads no envfile and uses fixed https://keryx.cc; metadata CSV protects formula-leading text. See source roles, UNRUN fixtures and original268 remaining gates.

evidence_procurement prepares private missing-evidence requests, compares signed provider quotations within an integer quote budget, checks separately signed delivered bytes and retains explicit supporting/refuting reviews. Exported packets can be restored and reassessed on web, API and CLI using the same contract. This proposal tool does not contact providers, fetch sources, execute benchmarks, purchase or reward anyone. Signed quotations confer no payout or source ownership authority; coverage/reviews remain declarations and payment is unobserved. See the workflow and remaining hosted/funded gates.

The source supports optional language: en|vi|pt|pt-BR|es|fr|zh-Hans|de|ja|ko|hi|id|ru|ar on ask_keryx for human-readable presentation only. It is consumed locally after the buyer returns: it is not added to the paid request/body/hash or forwarded as a model-language directive. Omission uses the original question's finite selector; explicit overrides disclose that actual answer language is unrecorded. Serving/output-limit notices share the API/remote MCP catalogue. The original answer/excerpts, structured result, custody/recovery path, payment/settlement authority text, canonical USDC amount and receipts stay unchanged. Package build/publication and actual delivery remain root-owned gates; no live call is required or executed by these source checks. See the exact shared contract and coordinator inputs.

For a new ordinary Arabic answer, put أجب بالعربية. or Answer in Arabic. in question; language: "ar" selects labels only. Stored answers retain their original text and require their existing recorded-language evidence.

App 0.27.60 is the next coordinated candidate for bounded ordinary optimizer replay, presentation-only research/bot labels and default-OFF deadline observation. The read-only idle signing correction joins this cohort after its own CI gates. Candidate identities are stdio/package/Registry 0.4.19, hosted MCP 0.3.17 and Desktop 0.4.17. The unchanged extension payload retains 0.1.4.2. Exact final-source CI, compiled package vectors, artifacts, publication and deployed readback remain required; these numbers are not publication or install proof.

Production app 0.27.59 /513c6fa6 was actually delivered on October12, with hosted MCP 0.3.16 and published npm/Registry 0.4.18. Its compiled idle signing fixture refused unrelated traffic and supplied no four-case acceptance; #274 stays open until the corrected source is delivered and observed. Zotero, privacy and deadline-observer activation retain their explicit holds. No new schema/backfill, schedule, payment or custody authority is granted. Dated notes below retain their original source and delivery evidence.

App0.27.58's iPhone header repair is scoped to the hosted web UI. It retains hosted MCP0.3.15 and stdio/package/Registry0.4.17 tool contracts and identities; existing artifacts keep their original source provenance. No new MCP package is required by this increment. See the surface audit and acceptance record.

The app0.27.57 coordinated candidate selects stdio/package and official Registry 0.4.17, with hosted MCP 0.3.15. Current main adds local draft_check and updates evidence_draft retained excerpts and author-written statements. Their actual runtime inputs changed, so published0.4.16 retains its earlier delivered source rather than representing this candidate. Both tools process caller text; they fetch no source, call no research/provider/storage/payment path, and confer no signing or private-history authority. Imported origin remains unverified.

Exact combined source/CI/build, the actual release bundle bound to the existing private-tool package fixture, initiali

Only the beginning is shown; read the rest in the repo