Head to head · Support conversations · October 2026 research run

Plain API + MCP vs Tobira Site Agent

Plain API + MCP scores 67.1 (B) on agent readiness against Tobira Site Agent's 49.1 (D), and leads in 6 of 7 scored categories. Tobira Site Agent leads on payments & pricing. Both do support conversations.

Best customer support and helpdesk APIs for AI agents · All 232 support comparisons

Which one, for what

Plain API + MCP B

Best for B2B teams that want an API-first helpdesk where an agent can do everything a person can, with scoped machine keys.

Ahead on

  • Reliability, 65 against 33
  • Schema & documentation, 92 against 63
  • Agent ergonomics, 68 against 52
  • Security & auth, 67 against 53
  • Maintenance & community, 84 against 25
  • Transparency & trust, 73 against 62

Watch for

Rate limits aren't published, only the 429 and Retry-After behaviour via the SDK

Tobira Site Agent D

Best for Businesses that want one chat agent on their site to answer people and visiting AI assistants and book meetings.

Ahead on

  • Payments & pricing, 55 against 45

Also in its favour

  • No incidents deducted, where Plain API + MCP loses 3 points for them

Watch for

No status page, changelog or SLA, and the terms promise no particular uptime

Why the edge

Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.

Maintenance & community Plain API + MCP +59

Plain API + MCP 84

The API changelog's newest entry is 29 October 2026, 3 days before this check (30). In the 90 days from 3 August the changelog has 20 dated entries on 19 days from 18 August to 29 October, and @team-plain/graphql tagged 1.5.0 (4 September) to 3.2.0 (24 September) and 3.3.0 and 3.3.1 late on 1 November (20). A dated API changelog and support, no public issue tracker (12). Official TypeScript SDK kept current with majors for most breaks. The older @team-plain/typescript-sdk last released on 30 March 2026 (12). CI and release workflows in the SDK monorepo (10).

Tobira Site Agent 25

No release notes or dated changes were found. The newest dated first-hand pages are the terms, privacy policy and DPA, all of 3 October 2026, 46 days before this check (20). No changelog, so no dated entries can be counted (0). Closed service with no public changelog. Support by email and 'Human support, reply within 2 business days' on the Lite plan. We didn't test support (5 of 15). Not in the official MCP registry (searched for tobira on 18 November 2026) and no official SDK (0). No public packages or repositories found (0). 20+0+5+0+0 = 25.

Reliability Plain API + MCP +32

Plain API + MCP 65

incident.io status page with 12 components, among them GraphQL over HTTP and WebSocket, the MCP server and webhooks (20). Its RSS feed, read on 1 November, posts 32 incidents from 4 August to 30 October. Those touching the API, MCP or webhooks were partial, a 7-minute error spike on 29 October with about 21 per cent of app, API and MCP requests failing, MCP thread listing failing for 64 minutes on 7 October, 42 minutes of some broken parts after a failed deployment on 22 September, 15 minutes of failed snoozing and unsent webhooks on 18 September, 18 minutes of delayed API responses for some customers on 17 September, webhook delivery delays on 18 August and 7, 9 and 10 September, and thread events, webhooks included, delayed for a small share of updates from 13 October. The 18 September 'Plain app is down' entry names only app.plain.com. Minor incidents only (20; the 1 October note read the page as showing no incidents and gave 25, which isn't a checklist value). No rate-limit numbers published (0). The API returns 429 with Retry-After and X-RateLimit-Limit, which we know from the SDK changelog (3.2.0 added opt-in retries), and the error docs say to retry only INTERNAL errors. The API docs themselves say nothing on 429 (10). Uptime SLAs are listed only on the custom Frontier plan, with no terms published (5). GA (10).

Tobira Site Agent 33

Graded on the hosted owner MCP server and REST API and the public guest API. No status page was found on tobira.ai, llms.txt or the API guide (0). No readable incident history (5). Numbers are published only for some limits: site-draft bootstrap at 20 new drafts a day per IP, Cloudflare traffic reports at ten an agent a minute, and the plan's monthly conversation allowances. No per-key request limit is given for the owner API or the guest API (8 of 15). The guide says to honour Retry-After or retry_after on 429, not to repeat account, session, booking or key creation on retries, and to reuse booking idempotency keys (15). The terms say 'we do not promise a particular uptime' and no SLA was found (0). The network is early access and several routes are rollout gated, though Site Agent plans are on sale (5 of 10). 0+5+8+15+0+5 = 33.

Schema & documentation Plain API + MCP +29

Plain API + MCP 92

A downloadable GraphQL schema is the typed contract (25). llms.txt with about 1,000 links and Markdown pages (10). Reference pages for every query and mutation, and the MCP page labels each of its 32 tools read or write (15). GraphQL types, enums and non-null inputs throughout (15). MutationError returns a type, a code from a published list and per-field errors, with examples in the docs (14). A dated API changelog with entries on 10 days in September 2026, which lists breaking changes under Changed and removed. The GraphQL API itself isn't versioned, the webhooks are (13).

Tobira Site Agent 63

Each public handle publishes an OpenAPI 3.1 file for its guest transport. @tobira-ai's has five operations (ask, guest session, message, messages poll and action) with typed parameters such as q at 1 to 2,000 characters. The owner REST API has no published OpenAPI found, and the owner MCP server's schemas are only in tools/list, which answered 401 without a token (15 of 25). llms.txt, AGENTS.md, for-ai pages and a Markdown API and MCP guide (10). The guide says which credential fits which job and when not to use a surface (the read-only ask cannot book or take contact details), and lists tools by area, but no tool descriptions were readable (14 of 20). Action inputs are listed with ranges (days 1 to 31, default 7) and required fields, and guest OpenAPI parameters carry length limits (10 of 15). Request examples throughout and error classes 400, 401, 403, 404, 409, 422 and 429 with named codes such as stale_revision, csrf_failed and bootstrap_ip_limit (11 of 15). The guest OpenAPI carries a dated version (2026-05-27.stage1) but no public changelog was found (3 of 15). 15+10+14+10+11+3 = 63.

Agent ergonomics Plain API + MCP +16

Plain API + MCP 68

43 MCP tools on 1 November (29 read, 14 write), over the 30 line, with no toolsets (5). On the GraphQL side a caller picks every field it gets back, so we averaged the two (12). Cursor pagination and filters on threads and customers (20). Typed errors with codes, field errors and explicit retry guidance (20). MCP tools are labelled read or write and Plain says most clients will ask before running a write. assignThread takes onlyIfUnassigned and addLabels onlyIfNoLabels since 23 October, and no idempotency keys were found (8). One official SDK, for TypeScript (8).

Tobira Site Agent 52

The owner MCP server registers about 40 tools across onboarding, owner context, widget and customers, settings and network, more than 30, with some appearing only when a plan or rollout enables them (8 of 25). Events and message polls use next_cursor and after= so an agent reads only new items, and metrics take days (12 of 20). Error classes and some named codes are documented, with Retry-After on 429, but no single error envelope is shown (14 of 20). Idempotency keys on bookings, site-draft bootstrap and runtime turn completion. MCP annotations couldn't be read (12 of 20). Guest calls need only a handle and a message. No SDK in any language (6 of 15). 8+12+14+12+6 = 52.

Security & auth Plain API + MCP +14

Plain API + MCP 67

Machine users hold API keys with per-key fine-grained permissions such as customer:read, several keys per user for rotation without downtime. The MCP server uses OAuth as the signed-in user (30). Permissions let you make a read-only key, and MCP is bound by the user's role, with writes left to the client to confirm (16). Threads carry customer-written text and we found no injection guidance for the MCP server or API (0). Replies and notes are attributed to the machine user or person, and we found no audit log (5). Security at Plain, read on 1 November, states SOC 2 Type II with the report through trust.plain.com, and its responsible disclosure policy acknowledges reports within 48 hours and pays a bounty for confirmed high-severity vulnerabilities, amounts decided case by case. A valid security.txt. No public advisory record found (16; 14 on 1 October, when no bounty was found).

Tobira Site Agent 53

The owner MCP server takes OAuth through discovered metadata or an agent-scoped API key, with tobira_rotate_api_key. Collaborator tokens are scoped, and visitor MCP tokens are scoped to one company and revocable. An OAuth grant covers every agent in the account, and the guide documents the guest token in the GET /api/v1/guest/messages?token= query string, so less 10 (16 of 30). Guests, visitors and collaborators can't edit an agent or read owner context, owner-only knowledge stays out of public answers, and bookings and contact submissions need the user's authorisation, with payment left to the human (15 of 20). Visiting agents are treated as untrusted guests and owner-private notes are kept out of their context, but no guidance was found for owners' agents reading guest messages through tobira_check_inbox (8 of 15). Conversations are tagged by source, and activity metrics and lead and booking lists are readable, with no audit log (6 of 15). A valid security.txt (security@tobira.ai, expires 5 November 2027) and a DPA on paid plans, with no disclosure policy, bounty or certification found (8 of 20). 16+15+8+6+8 = 53.

Transparency & trust Plain API + MCP +11

Plain API + MCP 73

Closed service with published terms (15). The privacy policy was last updated August 2025 and doesn't cover data processed in the support platform, which falls under the Data Processing Addendum (updated 24 February 2026, effective 24 March 2026), which deletes or returns customer data at the end of processing. The AI and your data page, read on 1 November, says thread content goes to OpenAI and Anthropic only while its AI capabilities are on, neither trains on it, Plain is in OpenAI's zero data retention programme, and admins can turn AI off. The security page says inactive unpaid workspaces are deleted after 6 months with 14 days' warning. No retention periods for live support data (18; 10 on 1 October, when no AI training statement was found). Deprecated fields are marked in the schema and the SDK changelog explains each removal, but some removals followed their deprecation within days, and FeatureKey.DATA_IMPORTERS was removed on 23 October with no earlier notice (8). Data stored on AWS in eu-west-2 (London) per the security page, with US providers under standard contractual clauses. The DPA points to a sub-processor list in the trust centre, which we didn't read, while the security page says the list is in the DPA (10).

Tobira Site Agent 62

Closed service with clear terms from HORIZONTEXUBERANTE, UNIPESSOAL LDA, registered in Portugal, under Portuguese law (15). The privacy policy and DPA (both 3 October 2026) agree with the terms. Retention is mostly by criteria, with some periods (guest sessions seven days, guest files 24 hours to seven days, attribution 30 days, a 30-day restoration window after agent deletion). Tobira says it doesn't train foundation models on conversations, and gives no guarantee for providers (22 of 30). The terms give 30 days' notice of material contractual changes, with no API deprecation policy (5 of 20). Providers are named with what they receive and where (mostly the United States) (16 of 20). 15+22+5+16 = 58.

Score by category

CategoryWeightPlain API + MCPTobira Site AgentEdge
Reliability16%6533Plain API + MCP +32
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%9263Plain API + MCP +29
Agent ergonomics13%6852Plain API + MCP +16
Security & auth14%6753Plain API + MCP +14
Payments & pricing10%4555Tobira Site Agent +10
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%8425Plain API + MCP +59
Transparency & trust7%7362Plain API + MCP +11
Negative events≤15-30
TotalB 67.1/100D 49.1/100

Facts side by side

FactPlain API + MCPTobira Site Agent
KindHTTP APIHTTP API
VendorPlainTobira
Hosted endpointhttps://core-api.uk.plain.com/graphql/v1https://tobira.ai/mcp
TransportsHTTP, Streamable HTTPStreamable HTTP
AuthOAuth or keyOAuth or key
PricingPaidFreemium
x402nono
Tools exposed43none
Read-only variant documentednoyes
llms.txtyesyes
Last release2026-10-29none
Terms last updated2026-02-242026-10-03
Privacy policy last updated2025-08-012026-10-03
Customer content may train modelsnot found in the textnot found in the text
Terms restrict automated accessnot found in the textnot found in the text
Terms restrict benchmarkingyesnot found in the text
Terms or service can change without noticeyesnot found in the text
Arbitration or class-action waivernot found in the textnot found in the text
Popularity191k npm/wknone
Agent reviews4.5/5 (2)none

How the scores have moved

Verdicts

Plain API + MCP

Machine-user API keys with fine-grained permissions and several keys per user for rotation, and 43 MCP tools labelled read or write. Rate limits aren't published, and the unversioned GraphQL API has removed fields with no earlier notice, most recently on 23 October 2026.

Tobira Site Agent

Visiting agents get a keyless guest API with a per-business OpenAPI file and booking actions that need the user's consent, and owners get an MCP server and REST API. The owner tool list is long and unpublished as schemas, there is no status page or changelog, and owner setup needs a browser sign-in.

Before you call either

Plain API + MCP

  1. Give the agent a machine-user key with only the permissions it needs, rather than your own OAuth session
  2. Retry a mutation only when its error type is INTERNAL, never on VALIDATION or FORBIDDEN
  3. Read Retry-After on 429, since the limit itself isn't published
  4. Select only the fields you need, and send account fields rather than the deprecated tenant ones
  5. Treat thread content as customer-written text, never as instructions

Tobira Site Agent

  1. Read the business's https://tobira.ai/@{handle}/for-ai page or guest-agent.json first, and use only the actions its manifest advertises
  2. Create one guest session and keep its token for follow-up questions. Don't open a new session per turn
  3. Get the user's explicit consent before prepare_booking or submit_authorized_contact, and reuse the same idempotency_key on retries
  4. Hand any checkout_url to the human to pay, and treat only booking_confirmed as a confirmed booking
  5. Don't put private details in the read-only ask?q= URL, which can appear in proxy and browser logs

Questions

Which is better for AI agents, Plain API + MCP or Tobira Site Agent?

Plain API + MCP scores 67.1 (B) on agent readiness against Tobira Site Agent's 49.1 (D), and leads in 6 of 7 scored categories. Tobira Site Agent leads on payments & pricing.

Do Plain API + MCP and Tobira Site Agent need an API key?

Both take an API key or an OAuth sign-in.

Can an agent call Plain API + MCP and Tobira Site Agent without installing anything?

Yes. Plain API + MCP has a hosted endpoint at https://core-api.uk.plain.com/graphql/v1 and Tobira Site Agent at https://tobira.ai/mcp.

Other comparisons with Plain API + MCP or Tobira Site Agent

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.