"Spring Boot" matching MCP tools:

• bookA

  Aurinia — book across venues via your AI assistantA

  REAL action: book at a network venue on the user's behalf. provider_id + item (the venue interprets it). Doctor: item={date,time}. Hostel: item={check_in,check_out,guests,promo_code?} — do NOT ask for name/contact: identity comes from a one-time Telegram consent. Spa: item={date,time,procedure}. Hostel: book returns a HOLD with payment_url; the booking is finalized by PAYMENT, not by this call. CONSENT: do NOT ask for approval up front — just call book. ONLY if the call returns consent_url/telegram.consent_url (linked=false = not yet authorized at this venue) do you show that link to the user for a one-tap approval in Telegram, then retry. If consent_url_native (tg://...) is present, try to open it on the user's device (open/xdg-open/Start-Process — launches Telegram directly if installed). BUT tg:// fails SILENTLY without Telegram — so ALWAYS also show the user consent_url (https): t.me opens the app or offers to install it. Open once, then retry. If already authorized (consent given earlier) book succeeds IMMEDIATELY, with no prompts. Never ask for consent when you already have it.

  ConnectorNo auth

• generate_harmonyA

  rnv-color-mcpA

  Generate a color harmony from a base color. base accepts a hex, CSS name, RNV brand name, or saved-palette reference (e.g. 'Spring line:2'). scheme is one of: complementary, analogous, triadic, split-complementary, tetradic (a.k.a. square), monochromatic, compound. Returns a list of hex colors. Read-only and deterministic: it derives the colors from the base and stores nothing, so it has no side effects and is safe to call repeatedly. Use to expand one base color into a related set; to blend existing colors into a single color use mix_colors, and to persist a set you like use save_palette.

  ConnectorNo auth

• generate_harmonyA

  rnv-color-mcpA

  Generate a color harmony from a base color. base accepts a hex, CSS name, RNV brand name, or saved-palette reference (e.g. 'Spring line:2'). scheme is one of: complementary, analogous, triadic, split-complementary, tetradic (a.k.a. square), monochromatic, compound. Returns a list of hex colors. Read-only and deterministic: it derives the colors from the base and stores nothing, so it has no side effects and is safe to call repeatedly. Use to expand one base color into a related set; to blend existing colors into a single color use mix_colors, and to persist a set you like use save_palette.

  ConnectorNo auth

• convert_timeA

  Timezone TruthA

  Convert a wall-clock time between two IANA timezones, and report the ways the conversion can be wrong. ALWAYS use this instead of doing the arithmetic. Timezone conversion looks like addition and is not: the mapping from a local time to an instant is not a function. - On the spring-forward date, an hour of local time DOES NOT EXIST. Asked to convert 02:30 on that date, a model will return a plausible timestamp for a time that never happens. This tool returns null and says why. - On the fall-back date, an hour happens TWICE. There are two correct answers an hour apart; this returns both rather than silently choosing. - The gap between two zones is not constant. Zones start and end daylight saving on different dates, so a cached offset is wrong for weeks each year. - Not all offsets are whole hours: India is +05:30, Nepal +05:45. Input: `time` must be YYYY-MM-DD HH:MM (or with T, and optional seconds) — free-form dates are refused rather than guessed. `from` and `to` must be full IANA names such as "America/New_York". ABBREVIATIONS ARE REJECTED, deliberately, even though most runtimes accept them. Node resolves "BST" to Bangladesh Standard Time (UTC+06:00) when nearly everyone writing it means British Summer Time (UTC+01:00) — a five-hour error that yields a perfectly plausible timestamp. Same for CST, IST, PST, and EST. Returns: the chosen UTC instant (null if the local time does not exist), every candidate instant, both zone readings with the offset and abbreviation that applied on THAT date, the difference between the zones, upcoming clock changes for both, and warnings. Check `warnings` for severity "error" before using the result.

  ConnectorNo auth

• query_power_price_ercot_v1A

  exascale.buildA

  Query verified ERCOT wholesale electricity prices — ERCOT's Day-Ahead Market Settlement Point Prices ($/MWh, EMIL NP4-190-CD), the price cleared the day before each operating day at every ERCOT (Texas) settlement point, served hourly. Returns cited prices for each (`settlement_point`, `delivery_date`, `hour_ending`): the per-hour `price_usd_per_mwh` in detail records, plus `avg_price_usd_per_mwh`, `min_price_usd_per_mwh`, and `max_price_usd_per_mwh` over the result scope. Each `settlement_point` is one of ERCOT's locations — a trading Hub (e.g. `HB_NORTH`, `HB_HOUSTON` — the regional benchmark prices), a Load Zone (e.g. `LZ_HOUSTON`), a Resource Node (one generator's connection point), or a DC-tie — and `settlement_point_type` carries ERCOT's OWN verbatim classification code (`HU`/`LZ`/`RN`/`LZ_DC` and finer codes) so an agent can tell a regional benchmark from a single-plant node. Filter or group by `settlement_point`, `settlement_point_type`, `hour_ending`, or `delivery_date`; filter a date window with `delivery_date_from` / `delivery_date_to`, or one day with `delivery_date`. Pass each parameter as a top-level key of `params` (flat — not nested). Example: `{"settlement_point": "HB_NORTH", "delivery_date": "2026-06-20", "group_by": ["hour_ending"]}` for the North hub's 24 hourly day-ahead prices; `{"delivery_date": "2026-06-20", "group_by": ["settlement_point_type"]}` for the average price by location type. With no date filter the result defaults to the latest delivery day with prices (it does not scan all history); served delivery coverage begins 2014-05-02 from ERCOT's official Data Access Portal archive. Historical settlement-point names remain exactly as published, and names absent from the current pinned type list have a null type rather than being rewritten. Returns JSON with citations and optional row-level records when `include_records` is true; every value carries `source`, `as_of` (the delivery day), and a `source_row` verifiable with get_source_evidence_v1. A price is INTENSIVE ($/MWh): it is AVERAGED, min'd, and max'd over a scope — NEVER summed (a "total price" is meaningless, so no sum is offered). An average across more than one settlement point (e.g. a hub and a resource node together) is indicative, not a settlement value — group_by `settlement_point` for the per-point series, or filter to one point/type. This is the DAY-AHEAD hourly market, NOT real-time / 5-minute prices. ERCOT's day-ahead settlement-point price is a TOTAL only — there is no energy/congestion/loss component split and no loss component, and none is synthesized. ERCOT's own hour-ending label (`01:00`..`24:00`) and `dst_flag` are carried verbatim (a day is 24 hours normally, 25 on the fall-back DST date with `02:00` repeated, 23 on spring-forward). A settlement point is an electrical/aggregate location, not a plant — ERCOT supplies no county or lat/lon, so the only geography anchor is `state` = TX. This is a PRICE ($/MWh) — not capacity (MW) or generation (MWh): for installed/operating capacity use query_power_capacity_v1, for electricity generated use query_power_generation_v1. ERCOT only — prices are NEVER blended, averaged, or compared across ISOs (each ISO's market design and redistribution license differ); this tool serves ERCOT's day-ahead settlement-point prices alone.

  ConnectorNo auth

• get_vps_statusA

  EQVPS — VPS and code sandboxes for AI agentsA

  Get full detail for one VPS: status, specs, live VM state/uptime, and access info (host, port, ready-to-paste `command`; `access.method` = ssh for Linux, rdp for Windows — port follows the plan: dedicated IP 22/3389, NAT 20000+vmid). ROOT PASSWORD (keyless VPS): call with `reveal: true` to retrieve it. It is shown ONCE, and only after the VM has finished booting — at that moment the password is applied to the live VM, so it is guaranteed to work. While the VM is still booting the response carries `password_pending: true` and NO password; just call again with `reveal: true` in ~30-60s (the one-time reveal is NOT spent until a password is actually returned — no reset or reboot needed). Without `reveal`, `password_available: true` means a password is waiting to be revealed. If you supplied an SSH key at order time, log in with that key (no password); `access.ssh_ready` (Linux) / `access.rdp_ready` (Windows) is a LIVE probe of the login port: on a fresh order the first boot applies OS updates, so SSH usually opens 1-5 minutes after the VM is running. While it is false the ready-to-paste `command` is withheld and `ready_note` + `retry_after_seconds` tell you to poll again — do not reset, reboot or reinstall during that window. Use `reset_password` ONLY if the password is lost. Poll this (without `reveal`) after `order_vps` to watch provisioning reach `active`. NAT plans (no dedicated IP): the SSH `port` shown is an EXTERNAL DNAT port forwarded to the VM internal port 22 (`internal_port` in the response). If you configure a host firewall, allow the INTERNAL port — `ufw allow OpenSSH` — NOT the external port, or `ufw enable` will lock SSH out. See `firewall_note` in the response. Requires: token. Typical: access.ssh_ready=false right after an order — not an error, poll again after retry_after_seconds.

  ConnectorNo auth

• set_projectA

  clipkitA

  Replace the entire project with the given source JSON, returning its project_id. This is the PRIMARY way to build: use it to create a composition or to add/change many elements at once. (To tweak a single element in an existing project, use edit_element / add_element / delete_element instead.) REVISION RULE: if you are iterating on a video that already exists in this conversation — the user asked for changes, you are fixing your own work — you MUST pass its project_id so the project is updated in place and keeps one editor link and its version history. Omitting project_id creates a NEW, unrelated project and strands the old one; only omit for a genuinely different video. The input is validated against the @clipkit/protocol before being accepted; invalid inputs return an error. Shape: { width, height, duration, frame_rate, output_format, background_color?, defaults? (house easing, e.g. a spring), fonts?, camera?, lights?, elements:[…] }; every element has a `type` plus base fields (id, layer, x, y, width, height, time, duration, opacity, rotation, animations, keyframe_animations) and type-specific fields. CKP/1.1: keyframe easings accept a spring object {type:"spring", duration?, bounce?}; channels take stagger {each, from, ease, split, seed} to spread across text units / set copies / group children; groups take layout {direction, gap, align} (stack — no hand-computed positions); text takes a count-up {expr, format}. For exact field names + types call get_schema (optionally with an element_type) — the runtime ignores unrecognized keys, and this tool flags any it does not recognize.

  Connector

  Destructive

  No auth

• utm_builderA

  Moltline Outbound EngineA

  Build a UTM-tagged URL for campaign tracking. FREE. Typical input {"url": "https://example.com/pricing", "source": "newsletter", "medium": "email", "campaign": "spring-launch"} returns {"tagged_url": "https://example.com/pricing?utm_source=newsletter& utm_medium=email&utm_campaign=spring-launch"}. Use to build one tagged tracking URL. Not for analyzing campaign results and not for anything that belongs in the message body (audit_copy). Errors: on invalid, missing, or malformed input this tool never raises a protocol error — it returns {"error": "<what is wrong and how to fix it>"} (for example {"error": "url must start with http(s)://"}). Every call is read-only and idempotent, so after correcting the input it is always safe to retry.

  ConnectorNo auth

• utm_builderA

  Moltline Outbound EngineA

  Build a UTM-tagged URL for campaign tracking. FREE. Typical input {"url": "https://example.com/pricing", "source": "newsletter", "medium": "email", "campaign": "spring-launch"} returns {"tagged_url": "https://example.com/pricing?utm_source=newsletter& utm_medium=email&utm_campaign=spring-launch"}. Use to build one tagged tracking URL. Not for analyzing campaign results and not for anything that belongs in the message body (audit_copy). Errors: on invalid, missing, or malformed input this tool never raises a protocol error — it returns {"error": "<what is wrong and how to fix it>"} (for example {"error": "url must start with http(s)://"}). Every call is read-only and idempotent, so after correcting the input it is always safe to retry.

  ConnectorNo auth

• set_app_secretA

  DockholdA

  Store a sensitive value (an API key, a token, a password, a connection string) for one app. Use this instead of set_app_variable whenever the value would be damaging if someone else read it. The value is encrypted at rest, is injected into the app's environment under the name you give, and the app restarts to pick it up. Storing a secret takes effect on the next boot, so call this before telling the user the app is ready. This tool never reads a value back: there is no way to retrieve a stored secret through this connection, only through the dashboard while signed in. Two things to tell the user before you call it. First, this token can write secrets to every app on their account, not only this one. Second, any value they give you has passed through your context, so prefer a value you read from a local file such as .env over one you asked them to type into the chat, and suggest they rotate anything that was pasted. If the vault key name is already used by a different app, this call is refused rather than overwriting it, and the message tells you what to pass as vault_key_name.

  ConnectorNo auth

• book-contractA

  Commerce for AgentsA

  Files a concierge switch request with the complete contract data in one call. Intended for autonomous agents that already hold the user's full profile, IBAN and a standing authorization; in a conversation with a person, start-booking opens a form in which the user enters these data. Commerce for Agents stores the request encrypted and an operator submits the contract application to the provider; the user receives a confirmation email at once and the provider's contract documents later. There is no self-service checkout link. Input: the `ref` of the chosen offer plus the contract data. Postal code and city come from the comparison; previousProvider, previousCustomerNumber and meterNumber are printed on the last energy bill. `consent: true` attests that the user explicitly agreed to (1) Commerce for Agents submitting the contract on their behalf, (2) the SEPA direct-debit mandate for the given IBAN and (3) the privacy policy at commerceforagents.com/privacy, either by an explicit yes in the conversation or by a standing authorization given to the agent. The response contains nextSteps for the user.

  Connector

  Destructive

  No auth

• deploy_virtual_machineA

  Cycle MCPA

  Deploy a virtual machine onto Cycle. Prefer containers (deploy_application) for ordinary workloads; choose a VM only for hard isolation requirements, a custom OS or kernel (custom modules, non-Linux), or legacy software that cannot be containerized. Platform rules this tool applies or checks: - The environment's cluster must contain a hypervisor-capable server. list_servers reports 'virtualization' per server; the response notes when no live server in the cluster confirms it. When hardware has to be provisioned for VMs, get_deployable_server_models reports 'hypervisor' per model (and takes hypervisor_only). - Image: exactly one of base_image or image_url. Call with NEITHER to fetch the live base-image catalogue (version identifiers, supported/UEFI flags) without deploying — do that first unless the user named an exact image; prefer versions marked supported. iPXE and external-volume image sources are not exposed here (a 'base' volume backed by a SAN volume is a different thing and IS supported). - Resources are explicit: ram is required, plus exactly one of cores or cpu_pin. - Storage: volumes MUST include the boot volume, identifier 'base'; creates without one are refused. Other volumes attach as RAW BLOCK DEVICES the guest must partition, format, and mount — remind the user. - Access: Cycle generates a root password at create. It is returned here but retrievable for only ~10 minutes, so relay it to the user promptly (afterwards reconfigure_virtual_machine sets a new one). SSH keys attach at provision time: ssh_keys references existing environment-scoped keys, new_ssh_keys creates and attaches them. Serial-over-SSH console access needs no VM networking: get_vm_console_access mints credentials for the user's own interactive session, and run_vm_command runs a single command and returns its output (in-guest setup like partitioning a volume goes through it). - Networking follows the container conventions: IPv6-ONLY private network, hostname defaults to the identifier, public defaults to 'disable', ports map like '443:443'. Point a domain at the VM afterwards with manage_dns_record (records can link to VMs). - Placement: constraints.node.tags.all/.any restricts which tagged servers may host the VM — same tag model as deploy_application; useful when only some servers are hypervisor-capable. Workflow: 1. If the user hasn't picked an image, call with no base_image/image_url to list the base images. 2. Call with preview:true — returns the exact create request (read-only lookups only; nothing is created) to confirm with the user. Never create without explicit confirmation. 3. Call again without preview. The VM is created and, unless start:false, started. The first boot downloads the disk image and can outlast wait_seconds; the job keeps running on Cycle (check list_virtual_machines or get_jobs). Retries are safe: identifier (default: slug of name) is the idempotency key — a repeat call refuses to create a duplicate and reports the existing VM. Pass a fresh identifier to deliberately create another alongside it.

  Connector

  Destructive

  API key

• migrate_instancesA

  Cycle MCPA

  Migrate a container's or virtual machine's instances between servers on Cycle, or revert a recent migration. Pass exactly one of container or virtual_machine. Cycle backs every VM with a container, so a VM migrates through that container's single instance (select it with all_instances:true or name it in targets); the response reports both the VM and its backing container. Cycle migrates instances across any infrastructure it manages — between servers, data centers, cloud providers, and on-prem hardware. Not every server is a valid target: containers carry tag restrictions and other constraints, so this tool only accepts destinations Cycle reports as compatible for the container. A migration is reversible: the original instance is retained until Cycle's purge window elapses (roughly 3 hours for stateful instances), during which action:"revert" restores it on its source server. Only retained source instances are revertable; running destination copies are skipped. Load balancer instances cannot be migrated. Workflow: 1. Call with preview:true and your selection. It makes NO changes and returns the selected instances — each with its current server (id and name) and whether it is stateful — plus, for migrate, the compatible destination servers. The plan comes from read-only lookups; Cycle does not validate it. 2. If the destination is unclear, present the compatible servers and let the user choose. With NO compatible servers migration is impossible — tell the user why (tag or infrastructure constraints). 3. Confirm the specific move with the user, then call again without preview. Never migrate or revert without explicit confirmation. Select instances with exactly one of: targets (specific instances, each optionally with its own destination so they can spread across servers), source_server ("move this container off nuc-bear"), or all_instances. A top-level destination_server is the default for selected instances without their own and is required with source_server or all_instances. copy_volumes applies to stateful instances and defaults to true so data is never silently dropped. A VM's local volumes — including its boot disk — are what it moves, so copy_volumes:false lands the VM on empty local storage and is almost never wanted. External (SAN) volumes are attached, not copied, and are unaffected either way. Asynchronous: submits one job per instance and returns without waiting (wait_seconds gives a short bounded wait for quick moves). Track the job_ids with get_jobs, then call again with preview:true to confirm each instance reports its destination server.

  Connector

  Destructive

  API key

• save_paletteA

  rnv-color-mcpA

  Persist a named color palette for later retrieval with get_palette or list_palettes. Each entry in colors accepts what every other tool accepts: a hex (#d2bc93), a CSS name (red), an RNV brand name (brand gold), or a saved-palette reference ('Spring line:2'); each is resolved and stored as normalized hex, and an unknown token refuses the whole save naming its position. Optional notes are stored as the palette's description. Author is recorded as RNVizion. The name is refused, with the reason, if it is an RNV brand name, a CSS color name, a 'css:' form, a hex literal, or contains ':' -- a palette resolves ahead of brand and CSS names, so such a name would redefine a color for every caller. This WRITES to the palette store and is the only tool here that does. Reusing an existing name overwrites that palette: save and update are the same call (an upsert), there is no separate update operation. Returns a `durable` flag: true if the palette reached durable storage (the HF Dataset) and will survive a restart, false if it saved to the local working copy only (which is lost on rebuild), with `durable_reason` naming the step that stopped it. Use when the user wants to keep a set of colors under a name for reuse across sessions, such as a brand or launch palette; to read a palette back use get_palette, and to see what already exists use list_palettes. The saved name can then be passed to mix_colors, convert_color, and generate_harmony as a palette reference.

  ConnectorNo auth

• get_sites_by_stateA

  USGS Water MonitoringA

  Find USGS water monitoring sites in a state. Returns a list of monitoring stations with their site numbers, names, and locations. Use site numbers with get_water_levels to retrieve data. Args: state: Two-letter US state abbreviation (e.g. 'CA', 'TX'). site_type: Type of monitoring site. 'ST' for stream/river, 'GW' for groundwater well, 'SP' for spring. limit: Maximum number of sites to return (default 50).

  ConnectorNo auth

• reset_passwordA

  EQVPS — VPS and code sandboxes for AI agentsA

  Rotate the root password. The new password is applied to the LIVE VM immediately via the guest-agent (no reboot) and works within ~10-30s; the OLD password stops working. REQUIRES confirm = exact hostname (from get_vps_status) or 'DELETE'. Agent must confirm with the human before calling. Use this ONLY if the password was lost — NOT because SSH was not instant on a fresh order (a new VM needs ~60s to boot; poll get_vps_status with reveal:true instead). New password is NOT in this response — read it ONCE via get_vps_status (reveal:true); also emailed. Requires: token + confirm (hostname or DELETE). Only if the password is lost.

  ConnectorNo auth

• save_paletteA

  rnv-color-mcpA

  Persist a named color palette for later retrieval with get_palette or list_palettes. Each entry in colors accepts what every other tool accepts: a hex (#d2bc93), a CSS name (red), an RNV brand name (brand gold), or a saved-palette reference ('Spring line:2'); each is resolved and stored as normalized hex, and an unknown token refuses the whole save naming its position. Optional notes are stored as the palette's description. Author is recorded as RNVizion. The name is refused, with the reason, if it is an RNV brand name, a CSS color name, a 'css:' form, a hex literal, or contains ':' -- a palette resolves ahead of brand and CSS names, so such a name would redefine a color for every caller. This WRITES to the palette store and is the only tool here that does. Reusing an existing name overwrites that palette: save and update are the same call (an upsert), there is no separate update operation. Returns a `durable` flag: true if the palette reached durable storage (the HF Dataset) and will survive a restart, false if it saved to the local working copy only (which is lost on rebuild), with `durable_reason` naming the step that stopped it. Use when the user wants to keep a set of colors under a name for reuse across sessions, such as a brand or launch palette; to read a palette back use get_palette, and to see what already exists use list_palettes. The saved name can then be passed to mix_colors, convert_color, and generate_harmony as a palette reference.

  ConnectorNo auth

• upgrade_to_clusterA

  mcpA

  THE WAY TO MAKE AN EXISTING APP HIGHLY AVAILABLE. If the user asks for high availability, redundancy, no single point of failure, surviving a host or VM failure, or 'stay up if a machine dies', this is the tool: point it at their running VM and pass high_availability:true. It needs no rebuild and no redeploy, and the URL does not change. Turns a single running instance into a horizontally-autoscaling cluster: snapshots the instance and puts your SOURCE VM behind a load balancer as its ALWAYS-ON BASELINE MEMBER, then an autoscaling group adds EXTRA members (booted from that snapshot) on CPU load and scales them back down to zero. FLAVOR — LEAD WITH A MEMORY-OPTIMIZED SMALL-DISK FLAVOR (m1.mem16/m1.mem32) for the cluster, and ideally have the SOURCE VM on one too: clustering snapshots the source VM's FULL disk, so DISK size (not RAM) drives how long the cluster takes to build and to add each member under load — a 40 GB mem flavor is several times faster to snapshot and scale than a 160 GB standard flavor (m1.xlarge) and costs less for the same RAM. If the source VM is on a big-disk flavor, RECOMMEND migrating the app onto a mem-flavor VM first (a root disk cannot be shrunk in place), even though it is extra work — it pays off on every scale-out, and a clustered app is stateless anyway (state in the managed DB) so the lean disk fits. Expect a DRASTICALLY better clustering experience on a mem flavor: far faster cluster-create, snapshots, and every scale-out. The user can override with any flavor. At rest ONLY your source VM serves — there is NO idle extra VM to pay for (the source VM is the cluster's minimum, so the floor is 0 extra members). Use it to LOAD-SCALE a stateless app tier while managed services hold state: it becomes highly-available UNDER LOAD (multiple members behind the LB), but at rest a SINGLE source VM serves — and that source VM is a plain VM, not an autoscaled member, so it is NOT auto-replaced if it fails while idle (only the autoscaled extra members are ASG-managed and self-healed). If you need always-on redundancy, keep the app under enough load to hold >=1 extra member, or use a separate always-on setup. BILLED — at rest it costs just your source VM (which you already run) plus the load balancer; under load it adds up to max_size EXTRA members at the member flavor (flavor_id), billed only while they run. In guided mode show the cost that way (now: source VM already running + the LB; under load: up to max_size x the member flavor) and get the user's explicit go first. redu automatically repoints the extra members from the old single-VM URL to the load-balancer URL across app config. It REFUSES a STATEFUL VM with 409 cluster_needs_stateless unless confirm_stateless:true. To have redu FIX a stateful VM for you instead of refusing, pass auto_restructure:true — for a single_vm Postgres it fully-automatically provisions a managed DB + migrates the data + repoints the members; for a compose-stack DB it provisions the matching managed DB (set restructure_engine, e.g. 'mysql'/'mariadb' for WordPress) and returns migration commands to run from the app VM. WordPress/WooCommerce is not generic autoscaling: managed DB alone is not enough because wp-content/uploads is file state. Use app_profile:'wordpress'/'woocommerce', cluster_media_mode:'media_space', and either media_space_id or create_media_space:true so all members mount the same uploads filesystem; otherwise the backend refuses with 409 cluster_needs_media_space. PUT THE CLUSTER ON THE SAME private network as the managed DB and media space. HA: cluster members are spread across DIFFERENT physical hosts automatically, and an autoscaled member that is destroyed is REBUILT AUTOMATICALLY in 1.5 to 5 minutes depending on how it failed with no action from you (the always-on source/hero VM is a plain VM and is NOT covered by that). CRITICAL for members: the app must start on EVERY boot (systemd unit or container restart policy) - if it only starts from a first-boot cloud-init script, a rebooted or resized member comes back with no app, silently never rejoins the load balancer, and the cluster quietly loses capacity with nothing reporting an error. Pass startup_command if the app does not already auto-start on boot, and have it bind its port only once it is genuinely ready to serve (the health check can only see whether the port is open). SEQUENCING - this catches people: the snapshot is taken IMMEDIATELY, and every member boots from it, so the source VM's app must already be RUNNING before you call this. Clustering a freshly-created VM whose cloud-init has not finished captures an image with no enabled service, and all members then come up ACTIVE while failing the load-balancer health check forever - a cluster that looks built and serves nothing. Verify the app answers on its port first (get_ssh_command, or just fetch the VM's URL). The snapshot upload can take several minutes; poll list_clusters until CREATE_COMPLETE.

  ConnectorNo auth

• attach_volumeA

  mcpA

  Attaches an EXISTING block volume to a running instance as an extra disk. This is how you finish a RESTORE after a VM is gone: restore_backup (volumeName) rebuilds the data into a new volume, then attach_volume puts that volume on a replacement instance — then SSH in (get_ssh_command) and mount it (`lsblk` to find it, usually /dev/vdb, then `mount /dev/vdb /mnt/...`). The volume must be 'available' (not already in-use — detach_volume it first) and in the same project. Attaching does NOT mount or boot from it: the instance keeps booting from its own root disk, the volume shows up as an extra block device. Poll list_volumes until status is 'in-use'.

  ConnectorNo auth

• bookA

  Aurinia XO MCPA

  Book beds for a date range at Aurinia XO — you do the booking, never hand the guest a self-service link. Do NOT ask the user to confirm before calling this — just call it; surface a consent link ONLY if the response comes back pending_consent (telegram.linked=false). If the agent is already linked it books immediately with no confirmation (consent is one-time). Do NOT ask the guest for name or contact; identity comes from a one-time Telegram consent. If the response is pending_consent (telegram.linked=false), telegram.consent_url is a one-tap Telegram approve link: OPEN it yourself (open/xdg-open/start) if you run on the user's device, otherwise present it as a single prominent 'Approve in Telegram' link — do NOT tell the guest to book manually. After they tap Approve, retry this call. Consent is one-time (valid ~90 days); later bookings skip it. Once confirmed it returns a HOLD with a payment_url. IMPORTANT: this does NOT finalize the booking — the stay is confirmed only when the guest pays via payment_url before hold_expires_at; otherwise the hold auto-expires. Pass promo_code (e.g. BOT5) for a discount. The door code is delivered to the guest in Telegram after payment (and via list_current_bookings).

  ConnectorNo auth

• attach_compute_volumeA

  BorealHostB

  Restore a detached volume onto a fresh instance — optionally a new type. Pass a different instance_type to move the same machine to different hardware: this is the closest thing to changing instance type that the provider allows, since it has no resize API at all. The restore runs at first boot and the machine reboots into itself, so the instance answers SSH BEFORE it is ready. Poll get_compute_volume and give it a few minutes (measured: ~6 min for a 10 GB GPU machine end to end). The machine keeps its SSH host keys, so your client will not warn about a changed host identity. Requires: API key with write scope. Args: volume_id: A volume in state "detached" ssh_public_key: Required — installed via cloud-init at boot instance_type: Defaults to the type it last ran on root_disk_size: GB. May grow, never shrink below the volume's size. Returns: Volume dict with state "attached" and a new instance. Errors: CONFLICT: Volume is not detached, or has no snapshot yet

  ConnectorNo auth