Skip to content

API Plan Finder MCP server

io.github.scraping-api-plans/web-data-plan-finderscraping-api-plansstreamable-httpv0.1.6Scanned

Best API plans + report on 46 conditions for scraping, SERP, search, browser, email, TTS, STT, docs

12 tools: 1 do something that can't be undone, 8 change things and 3 only read.

Tools: 0 delete, 1 irreversible, 8 write, 3 read.
  • Reads private data · 2
  • Reads text others write · 2
  • Can send data out · 1
  • 3 findings

Tools listed without sign-in. Remote posture

With Anlyon

1 tool here does something that can't be undone. Govern it.

  • Anlyon lists the tools and adopts none.
  • Each tool you adopt gets approvals and receipts.
  • Your agent never holds the key.
Govern this server →

A tool your agent calls directly is not routed through Anlyon and is not governed by it. Anlyon Radar only lists and labels, and these labels are the same either way.

What the API Plan Finder MCP server can do

The API Plan Finder MCP server is published by the GitHub account scraping-api-plans. It has 12 tools. It can do things that cannot be undone, such as sending, paying or publishing (send_feedback) and create or change data (8 tools: check_conditions, cost_scenarios, share_ranking, relax_conditions and 4 more), but cannot delete data. 3 tools only read.

2 tools read private data (get_analysis_report and search); 2 tools bring in text other people can write (get_analysis_report and search); 1 tool can send data out of your system (send_feedback). On its own it has all three parts of a prompt-injection exfiltration path: private data, text others write, and a way out.

The fixed checks matched 3 findings (long-description and annotation-conflict). Each one comes with the text it matched.

Its remote endpoint lists its tools without sign-in, and its tools changed 2 times since Anlyon Radar first saw them.

Last scanned Nov 8, 2026, version 0.1.6.

Tools

Grouped by what each tool does, most consequential first. Open one for its reasons and full description; descriptions are the publisher's own words.

12 tools

Can't be undone 1Claude Code: ask
  • send_feedbackUse this when: after you gave the person an API plan or ranking from this service, they say what they think of it (useful or not, a price or fact that looks wrong, something missin…Can send data outNo findings

    Why it is in this group

    • The name "send_feedback" says "send": a word for sending, and what is sent cannot be taken back.

    What it can reach

    • outboundThe name "send_feedback" says "send": it sends or publishes something to other people.
    Description, as the server wrote it
    Use this when: after you gave the person an API plan or ranking from this service, they say what they think of it (useful or not, a price or fact that looks wrong, something missing, the plan they chose), or you work for a provider listed here and a fact about your own plans looks wrong, missing or outdated (include the official page in source_url). Pass the gist in one call: from, kind, and optionally the offer_id, what it is about, and a short message. Do not use this to ask for a field, provider, condition or plan to be covered (use request_coverage), and do not include a name, e-mail address, link or other personal information. Nothing is returned beyond the receipt, and nothing sent here changes the published facts: they change only after we re-check the official page. Example: from user_assistant, kind wrong_fact, about price, message "the page says 49 a month".

    Annotations the server declares

    • title: Pass on what the person thought of a plan or ranking
    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: false
    • openWorldHint: false
Changes things 8Claude Code: ask
  • canvas_doUse this when: you have opened a canvas with canvas_open, or a search answer gave you a canvas_id for its analysis_url page (that page is a canvas too: once the person has opened i…No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Use this when: you have opened a canvas with canvas_open, or a search answer gave you a canvas_id for its analysis_url page (that page is a canvas too: once the person has opened it and tapped it you can change it), and want to show the user what you are explaining. One action per call (unknown or empty args are refused, not ignored; the applied field of the result says what changed): send action and args, e.g. {canvas_id, action: "sort", args: {by: "cost"}}, and make a second call for the next. Actions: filter {only_full, billing, providers}; exclude {offer_ids}; sort {by: cost|name|provider; on a search answer page also score, its default}; compare {offer_ids, 2 to 4} puts plans side by side and charts their cost by volume; highlight {offer_ids, note}; explain {text} adds your explanation to the canvas, marked as yours; what_if {change: search fields} asks the engine again with changed conditions (a budget goes here as max_monthly_cost_usd; a field set to null is removed; at most 6 per canvas); chart {kind: cost_by_volume|none}; reset. To narrow a search answer or add a condition, call search again with this canvas_id (the person keeps the same page). Returns the new summary. Read each shown plan's fails and not_stated before using what_if.

    Annotations the server declares

    • title: Change what the canvas shows
    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: false
    • openWorldHint: false
  • canvas_openUse this when: the user is choosing an API plan and you want to show them the comparison on a page they can open, which you then steer with canvas_do.No findings

    Why it is in this group

    • The name "canvas_open" says "open": a word for a change.
    Description, as the server wrote it
    Use this when: the user is choosing an API plan and you want to show them the comparison on a page they can open, which you then steer with canvas_do. Before calling, ask the user for everything the comparison needs: what the API is for, the monthly volume, the budget, must-have conditions (billing, regions, concurrency, compliance) and what would be nice. Put every must-have you can into conditions (the same fields as search; budget as max_monthly_cost_usd, billing as billing_types); only conditions in these fields are checked. Returns canvas_url, the link to give the user, and a short summary: the shown plans with whether each meets every condition, where its cost sits among the shown plans (cost_band), the conditions it fails (fails) and the ones its provider does not state (not_stated). The exact costs, workings, sources and plan buttons are on the canvas only.

    Annotations the server declares

    • title: Open an analysis canvas for the user
    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: false
    • openWorldHint: false
  • check_conditionsCall this right before choosing or buying an API plan: it returns, for each plan, whether it meets every condition, the monthly cost, and the official sentence behind each verdict.2 findings

    Why it is in this group

    • The description's first verb is "call": a word for a change.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    Findings

    • Info

      The description is 2,277 characters long, over the 2,000 this check looks for.

      Matched text

      description ends: "...ition. Example: 2 plans by id against 100,000 emails a month with a dedicated IP."

      What this check looks for
    • Warn

      The server marks "check_conditions" read-only (readOnlyHint: true), and its own name, description or input give the class "write".

      Matched text

      annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":true,"openWorldHint":false}; The description's first verb is "call": a word for a change.

      What this check looks for
    Description, as the server wrote it
    Call this right before choosing or buying an API plan: it returns, for each plan, whether it meets every condition, the monthly cost, and the official sentence behind each verdict. Use this when: the user names specific plans (by name, id or pricing-page URL) and wants to know whether they meet the workload and conditions, or which plan of the same product is the cheapest that does. Do not use this when: the user has no plans in mind yet (use search). Call it right before you choose or buy a tool or plan; format "brief" returns a short answer. Check specific plans, given by plan id, plan name or pricing-page URL, against the workload and conditions; up to 20 plans in one call are judged against the same conditions. Returns met / not met / unknown per condition with the official source sentence, which plans and billing options each value applies to, and a purchase link. Each plan's result also carries cheapest_in_product (the cheapest plan of the same product that meets every required condition, with its monthly cost, or null if none does), cheapest_elsewhere (the cheapest plan of another provider in the same field that meets every required condition, with the official sentence for each condition the checked plan missed) and, with two or more plans, differs (the conditions where the checked plans split, per plan, with the monthly cost gap) and, with exactly two plans, same (the conditions on which both have the same verdict). What each provider's official pages do not settle is on /vendors/<provider_id>. The optional question is the user's own words; it is read only for words that suggest a condition the request does not carry (returned as request_may_omit) and is not judged or stored. An ambiguous name or URL returns the candidate plans instead of a guess. If you already have candidate plans or providers (for example from a web search), pass them as providers (search) or offer_ids (check_conditions) so that only those are judged. Prices are as of the as_of date of the result; call again right before buying; when you quote a result to the user, give its result_page. Each plan's link url is its sign-up page; the evidence sources are the pages that state a condition. Example: 2 plans by id against 100,000 emails a month with a dedicated IP.

    Annotations the server declares

    • title: Check plans against conditions
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • cost_scenariosUse this when: the user asks how the cost changes as the workload grows or shifts, or at what volume another plan becomes the cheapest.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
    Description, as the server wrote it
    Use this when: the user asks how the cost changes as the workload grows or shifts, or at what volume another plan becomes the cheapest. Do not use this when: the workload is fixed (use search). Show how API plan costs change when the monthly workload changes, for example twice the pages or a larger share of JavaScript pages. Along the path from the current to a target workload it returns the cheapest fully matching plan at both ends, each workload where the cheapest plan changes, and for tracked plans where the cost first steps up, where the plan can no longer cover the volume and where the budget is exceeded. Costs follow published pricing rules (whole packs of extra credits, per-period pack limits, per-connection browser billing, monthly minimums) with official source sentences. Example: 10,000 searches a month growing to 200,000.

    Annotations the server declares

    • title: Plan costs as the workload changes
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • relax_conditionsUse this when: search finds no plan or too few plans and the user wants to know which condition to loosen and by how much.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
    Description, as the server wrote it
    Use this when: search finds no plan or too few plans and the user wants to know which condition to loosen and by how much. Do not use this when: plans already match (use search or check_conditions). When no API plan, or too few API plans, meet every required condition: for each condition the user can change, returns the smallest change that lets at least one plan meet everything (for example the monthly budget needed, the concurrency available, or one exit-IP country to drop) and the plans it admits with their monthly cost. Also lists plans that fail only because a condition is not stated on the official pages, with the pages checked and the check date, so the user can confirm with the provider. Example: no plan meets 2,000,000 searches a month under $100.

    Annotations the server declares

    • title: Smallest condition changes that admit plans
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • request_coverageUse this when: a search or check of API plans came back empty or mostly unknown, or the user asks about a field, provider, condition or plan this service does not cover, and you wa…No findings

    Why it is in this group

    • The name "request_coverage" says "request": a word for a change.
    Description, as the server wrote it
    Use this when: a search or check of API plans came back empty or mostly unknown, or the user asks about a field, provider, condition or plan this service does not cover, and you want it to be considered. Do not use this when: plans were found (use them), or to ask a question (nothing is answered). Leave one short line: kind, name and an optional reason. No personal information: it takes no name, e-mail or link and refuses them. It is kept in the server log only and has no reply. Example: 1 search with 3 required conditions came back with 0 plans, and the user needs a condition that is not recorded.

    Annotations the server declares

    • title: Ask for a field, provider, condition or plan to be covered
    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: false
    • openWorldHint: false
  • share_rankingUse this when: the user wants the ranking of API plans for their conditions, or a link they can open or share, and you do not need every plan in the answer.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
    Description, as the server wrote it
    Use this when: the user wants the ranking of API plans for their conditions, or a link they can open or share, and you do not need every plan in the answer. Do not use this when: you need the plans and their sentences in the answer (use search with detail or format full) or the user named the plans (use check_conditions). Returns the one plan that leads the ranking for the conditions (a plan that meets every condition, or the closest when none does), a confidence (its score out of 99), how many other plans also meet every condition, and ranking_url: the permanent address of a page that ranks every plan for these conditions with the official sentence for each. Confidence is the plan's score out of 99, the same as in the search answer; it is not a probability. The conditions are those of search. Example: 50,000 SERP searches a month with an official MCP server; the person gets one pick and a page ranking every plan.

    Annotations the server declares

    • title: Ranking page and one pick for a person's conditions
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • verify_answerUse this when: you have drafted an answer that names API plans, costs or whether they meet the user's conditions, before you give it.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
    Description, as the server wrote it
    Use this when: you have drafted an answer that names API plans, costs or whether they meet the user's conditions, before you give it. Do not use this when: you have not looked up any plan yet (use search). Check what a drafted answer says about API plans against the recorded official facts, using the user's conditions as given in request. For each plan claim (plan id, name or pricing-page URL, with the stated billing, monthly cost, whether it meets every condition, single condition verdicts and recorded values) it returns conflicts with the official source sentences, the check date and a one-line fix: a plan not in the catalog, unmet or unknown conditions stated as met, met stated as unmet, a condition verdict, a cost outside the computed cost or range, or a value that differs. It also lists verdicts that rest on a derived value, official pages that disagree or a two-value price range (state the condition), plans it could not check (an ambiguous name returns the candidates), the cheapest plan that meets every condition when the answer does not name it, and words of the question that suggest a condition the request does not carry (the optional question is the user's own words; it is read only for those words and is not judged or stored). Example: a drafted answer says 1 plan costs $49 a month for 100,000 emails.

    Annotations the server declares

    • title: Check an answer's plan claims against the recorded facts
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
Only reads 3Show 3 read-only tools
  • get_analysis_reportYour reply to the person must state the plan, its score (N of 99) and the analysis link — the score is the point of this tool.Reads private dataReads text others writeNo findings

    Why it is in this group

    • The name "get_analysis_report" says "get": a word for reading.

    What it can reach

    • private-dataThe input names a table (input.needs_table_extraction), and the tool reads: it reads databases.
    • untrusted-inputThe description's first sentence says "reply": it reads comments, which other people can write.
    Description, as the server wrote it
    Your reply to the person must state the plan, its score (N of 99) and the analysis link — the score is the point of this tool. When the person says something is missing, wrong or not useful (for example a quality number they need), send it with send_feedback (from: user_assistant) — that is how this service learns what to add. Call this first when a person is choosing or buying an API plan (web scraping, web search or SERP, a hosted headless browser, transactional email, text-to-speech, speech-to-text, document parsing): pass question, the person's own words in full detail. Use this when: a person describes what they need from an API in these fields (a workload, a budget, needed features) or asks which plan to pick. Do not use this when: you only need the recorded conditions of one named plan (use get_details) or to check plans the person already named (use check_conditions). The server reads the kind of API, the monthly volume and the conditions from the words (read_from_question shows each field it read and the words that gave it); every search field you also pass is optional and, when you give one, yours is used. It returns the plan that fits most closely with its score out of 99, why, the analysis page the person opens, and how to try it for free. When the kind of API or the monthly volume is missing, or the words do not settle it (two kinds of API, two different figures, a correction), it returns no score and ask_back, and read_from_question marks those items asked with no value: put those questions to the person, then call again with their answers added to the question. Example: "We record about 120 hours of customer support calls a month, need speaker labels and can spend $150 a month" returns a plan, its score out of 99 and the analysis link.

    Annotations the server declares

    • title: Analysis report for what a person needs
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • get_detailsUse this when: you need every recorded condition, pricing rule and source sentence of one plan or product found by search or named by the user.No findings

    Why it is in this group

    • The name "get_details" says "get": a word for reading.
    Description, as the server wrote it
    Use this when: you need every recorded condition, pricing rule and source sentence of one plan or product found by search or named by the user. Do not use this when: you want to compare plans against a workload (use search or check_conditions). Get every recorded condition for one API plan or product: value, which plans it applies to, exceptions, official source sentence, check date and purchase link. A product too long for one response comes with one source sentence per fact. recent_changes lists the facts of the plan or product that changed in the last 30 days, with the previous and the new value and sentence and the date, and the facts whose sentence the daily re-check could not find again; it is left out when there are none. Each plan's link url is its sign-up page; the evidence sources are the pages that state a condition. Example: every recorded condition of 1 plan found by search.

    Annotations the server declares

    • title: Get recorded conditions of a plan or product
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false

This is not a safety rating, and no finding is not a safety guarantee. The groups and findings come from fixed checks of the tool definitions the server returned on the date shown, not from its code or what it does when called. How the checks work.

Exfil surface

Which tools read private data, which read text someone else can write, and which can send data out. One of each is a path a prompt injection can use.

4 paths in this server alone from private data, through text someone else wrote, to a way out. Denying 1 outbound tool breaks every one.

Reads private data2
Reads files, mail, messages, databases, secrets, calendars or repositories.
Reads text others write2
Brings in text someone else can write: web pages, inbound email, messages, issues and comments.
Can send data out1
Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.
deny to break every path
{
  "permissions": {
    "deny": [
      "mcp__web-data-plan-finder__send_feedback"
    ]
  }
}

Other servers in the same agent add their own tools to these paths. Audit your whole setup to see them together.

Remote posture

What anyone on the internet sees at the remote endpoint: an anonymous initialize and tools/list, and the public OAuth metadata.

Tools listed without sign-in

Looked at Nov 20, 2026, 19:23 UTC, unchanged since Nov 13, 2026. No credentials were sent and no tool was called. What is looked at.

Reachable
yes
HTTP status of an anonymous initialize
200
tools/list served without credentials
yes
Tools that change things, listed without credentials
9
Protocol version
2025-06-18
WWW-Authenticate on 401
not observed
resource_metadata in WWW-Authenticate
not observed
RFC 9728 protected resource metadata
not observed
Authorization servers
not observed
Legacy HTTP+SSE transport
no
Dynamic client registration
not observed
Client ID metadata documents
not observed
  • HighThe whole server

    The endpoint served tools/list without credentials, and the list has 9 tools that change things (1 irreversible, 8 write). Whether a call also works without credentials was not tested: tools are never called.

    Matched text

    https://plans.intoperson.com/mcp: tools/list returned 13 tools to an anonymous client, including send_feedback (irreversible), check_conditions (write), cost_scenarios (write), share_ranking (write), relax_conditions (write) and 4 more

    What this check looks for

Posture changes

  1. Nov 13, 2026
    • Tools that change things, listed without credentials: 8 to 9
  2. Nov 6, 2026
    • Tools that change things, listed without credentials: 6 to 8
  3. Oct 29, 2026
    • Tools that change things, listed without credentials: 5 to 6

Findings

Matches of fixed checks, with the text they matched.

1 warn2 info
  • The server marks "check_conditions" read-only (readOnlyHint: true), and its own name, description or input give the class "write".

    Matched text

    annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":true,"openWorldHint":false}; The description's first verb is "call": a word for a change.

    What this check looks for
  • Infosearch

    The description is 4,231 characters long, over the 2,000 this check looks for.

    Matched text

    description ends: "...a month in GB under $100; 300 hours of Korean audio a month with speaker labels."

    What this check looks for
  • The description is 2,277 characters long, over the 2,000 this check looks for.

    Matched text

    description ends: "...ition. Example: 2 plans by id against 100,000 emails a month with a dedicated IP."

    What this check looks for

History

Changes to its tool definitions and to its remote endpoint, newest first.

  1. PostureRemote posture changed. Tools that change things, listed without credentials: 8 -> 9.View diff
    • Tools that change things, listed without credentials: 8 to 9
  2. Tools5 added, 6 changed.+5 added~6 changedView diff
  3. PostureRemote posture changed. Tools that change things, listed without credentials: 6 -> 8.View diff
    • Tools that change things, listed without credentials: 6 to 8
  4. PostureRemote posture changed. Tools that change things, listed without credentials: 5 -> 6.View diff
    • Tools that change things, listed without credentials: 5 to 6
  5. Tools1 added, 4 changed.+1 added~4 changedView diff

Install

Pinned to version 0.1.6. Values in angle brackets are yours to fill in.

claude mcp add --transport http web-data-plan-finder https://plans.intoperson.com/mcp
claude mcp add --transport http web-data-plan-finder https://plans.intoperson.com/mcp

Run in a terminal.

Claude Code permissions

Allow the reads, ask before a change or a way out, deny the deletes. Paste into .claude/settings.json.

.claude/settings.json
{
  "permissions": {
    "allow": [
      "mcp__web-data-plan-finder__get_analysis_report",
      "mcp__web-data-plan-finder__search",
      "mcp__web-data-plan-finder__get_details"
    ],
    "ask": [
      "mcp__web-data-plan-finder__check_conditions",
      "mcp__web-data-plan-finder__cost_scenarios",
      "mcp__web-data-plan-finder__share_ranking",
      "mcp__web-data-plan-finder__relax_conditions",
      "mcp__web-data-plan-finder__request_coverage",
      "mcp__web-data-plan-finder__send_feedback",
      "mcp__web-data-plan-finder__verify_answer",
      "mcp__web-data-plan-finder__canvas_open",
      "mcp__web-data-plan-finder__canvas_do"
    ],
    "deny": []
  }
}

Tools that only read are allowed, tools that change things ask first, tools that delete or cancel are denied. A read that can send data out (it takes a URL, say) asks first too. It assumes the server is installed as web-data-plan-finder. Review it before you use it.

Capability tags behind 3 rules
  • mcp__web-data-plan-finder__get_analysis_report private-data, untrusted-input
  • mcp__web-data-plan-finder__search private-data, untrusted-input
  • mcp__web-data-plan-finder__send_feedback outbound

Pin

Keep the decision: fail CI when the tools change, follow every change, or show a badge.

Pin in CI

Fail a build when the tools you reviewed are not the tools the server serves today.

curl -s "https://radar.anlyon.com/api/v1/servers/io.github.scraping-api-plans/web-data-plan-finder/verify?hash=aff7a3bc800bd53bcfc019b504f1c42ce545d939906d90fbba9ea62f51138950" | jq -e '.changed == false'

Pinned hash aff7a3bc800b, from Nov 8, 2026, 12:18 UTC.

Watch

Get told when its tools change: a new tool, a new description, a new class.

https://radar.anlyon.com/servers/io.github.scraping-api-plans/web-data-plan-finder/changes.xml

An Atom feed. Works in any feed reader or Slack's RSS app.

Badge

Anlyon Radar badge for io.github.scraping-api-plans/web-data-plan-finder
[![Anlyon Radar](https://radar.anlyon.com/badge/io.github.scraping-api-plans/web-data-plan-finder.svg)](https://radar.anlyon.com/servers/io.github.scraping-api-plans/web-data-plan-finder)

Source: MCP Registry. Synced Nov 8, 2026, 12:17 UTC.

API Plan Finder MCP server: questions and answers

Answered from the scan above, not from the publisher's description.

Is the API Plan Finder MCP server safe?

No scan can promise that. What its definitions show: 12 tools, 9 of which can change, send or delete something, and 3 findings from fixed checks. Allow the reads, ask before each change, and deny the deletes; the permissions block on this page does that.

Can the API Plan Finder MCP server delete data?

No tool is classed as destructive: going by their definitions, none of them deletes, cancels or revokes anything.

Can the API Plan Finder MCP server do things that cannot be undone?

Yes. 1 tool can send, pay, publish or merge: send_feedback.

What tools does the API Plan Finder MCP server have?

12 tools: 1 irreversible, 8 write, 3 read. send_feedback, check_conditions, cost_scenarios, share_ranking, relax_conditions, request_coverage, verify_answer, canvas_open and 4 more.

Does the API Plan Finder MCP server need sign-in or OAuth?

Its remote endpoint listed its tools to an anonymous client, without sign-in.

Does the API Plan Finder MCP server send data to outside services?

1 tool can carry data out (it sends, posts or publishes, or takes a recipient, channel or URL): send_feedback.

Has the API Plan Finder MCP server changed its tools?

Yes, 2 times since Anlyon Radar started tracking it, most recently on Nov 8, 2026. Every change is on this page and in its Atom feed.