Pod

Available as Markdown and JSON. Pod is also available over MCP.

MCP Servers by devino.ca

Every MCP server published by devino.ca that Pod indexes.

4 servers, ordered by decision readiness and adoption — GitHub stars and npm downloads where available, then reported issue volume. Pod has connected to 4 of them; the rest are publisher-reported and unverified.

BioFlow

BioFlow is a link-in-bio page with a built-in AI agent. The BioFlow connector lets Claude manage that page end-to-end: list and inspect your bio pages, edit the draft (add, edit, remove and reorder blocks — links, images, products, newsletter signups, booking embeds and more), create new pages, pull analytics (views, clicks, CTR, top links, referrers, tip revenue), and read captured leads and uploaded files. Twelve tools over Streamable HTTP: five read-only, five draft writes, and two publish tools.

Safety is built in. Every connection goes through OAuth with per-scope consent, and a tool whose scope you did not grant is never registered for the session, so Claude cannot see it or call it. Grants are listed and revocable in Settings > Connected AI apps. Draft edits never touch your live page. Publishing is double-gated: it requires a workspace-level "dangerous operations" toggle that is OFF by default, and every publish is two-step — Claude first receives a preview of exactly what would go live plus a short-lived confirmation token, and only a second, confirmed call commits it. Optimistic-concurrency checks refuse writes when the draft changed under Claude's feet, and every tool call is audited and visible in your dashboard.

The connector sells nothing in-chat, generates no images, video or audio, and serves no ads. Works on every BioFlow plan; usage rides your plan's normal limits.

12 tools reported — Live, but requires authorization before it will list tools. Publisher lists 12 tools.

doDomain

doDomain is domain-connect-as-a-service: it gives a product a guided flow for pointing a customer's own domain at that product, and handles the DNS records and verification behind it. The doDomain connector gives Claude seven tools over your own team's state. Claude can pre-flight a domain before anyone touches DNS - which provider manages it, the registrable zone, and which connect tier it qualifies for (one-click OAuth, Domain Connect, or guided manual). It can list the apps on your team with their ids, widget keys and sandbox flags; list your existing domain connections and their live DNS health, filtered by app or domain; start a new connect session and return the records that need applying; fetch the state of a session you already started; check a session's expected records against authoritative nameservers, with a per-record breakdown of what is present, propagating or absent; and queue an on-demand recheck of a connection that may have drifted.

Access runs through OAuth 2.1 with PKCE and dynamic client registration. Scopes are granted per capability on the consent screen - separate read and write scopes for domains, apps, connections and sessions - and a tool whose scope you did not grant is never offered to Claude. Revoke a connected assistant any time from Connected Apps in the dashboard. Tool responses follow a narrow contract: secret API keys are never returned, and doDomain's billing surface is not reachable over the connector at all. Three of the seven tools read the public DNS system and are marked open-world for that reason; none changes anything outside your own doDomain team.

Two things to know: creating a connect session counts against your plan's monthly connection quota, and verification checks are rate limited, so a repeated recheck is told to wait rather than silently retried. The connector moves no money, generates no images, video or audio, and serves no ads. Documentation is at dodomain.io/docs/connecting-ai-assistants.

7 tools reported — Live, but requires authorization before it will list tools. Publisher lists 7 tools.

GetItDone

GetItDone is a team task and workspace manager: workspaces, projects, tasks with rich-text notes and attachments, recurring occurrences, and dependency chains showing what blocks what. This connector lets Claude work inside your workspaces from your own account - list the workspaces you belong to and the projects in them, browse and filter tasks by status and priority, open a task in full with its notes, attachments and dependency chain, pull still frames from a video already attached to a task, create tasks, update them, tick off a recurring occurrence, link one task as a dependency of another, archive a task, and search GetItDone's own developer documentation.

Eleven tools over Streamable HTTP: six read-only and five writes. Archive is the only removal - nothing here permanently deletes a task, and a dependency link can be unlinked again, so no tool on this surface is marked destructive. Every tool touches only the workspaces of the signed-in account, so all eleven are marked openWorldHint false. You connect by signing in to your existing GetItDone account over OAuth 2.1 with PKCE. Three scopes are granted separately at the consent screen: reading workspaces, reading tasks, and writing tasks. A tool whose scope you did not grant is never registered, so it does not appear in the tool list at all; calling it anyway returns a structured SCOPE_MISSING denial naming the scope you would need. Review and revoke any connected assistant from the Connected apps tab in your GetItDone settings.

Attachment and video-frame links handed to Claude are short-lived presigned URLs to your own workspace files, not permanent public ones - they expire, so Claude re-fetches rather than caching. The connector processes no payments, generates no images, video or audio, and serves no ads: frame extraction reads a file you already uploaded, it synthesises nothing. Works on any GetItDone account that belongs to at least one workspace.

11 tools reported — Live, but requires authorization before it will list tools. Publisher lists 11 tools.

Uptimely

Uptimely is an uptime-monitoring and incident-management platform. This connector gives Claude read access to your monitors and their status history and targets, your incidents, alerts, maintenance windows, on-call schedules, status pages and telemetry services - and, once a project owner enables AI write operations, the ability to create a monitor, run an on-demand probe, declare an incident, change incident or alert state, save a postmortem and create an alert.

Every write tool stays inert until a project owner turns on "Allow AI write operations" in Settings > API Keys, and it is OFF by default - because some of these tools page real people or change what a public status page shows. While the toggle is off, a write tool returns a structured AI_WRITE_OPS_DISABLED result carrying a deep link to that exact settings page, and writes nothing. Six write tools dispatch outside Uptimely: a probe run hits a real URL, and declaring an incident or changing incident or alert state can notify real people. Those six carry openWorldHint in their metadata, so clients can prompt before running them. Nothing on this surface deletes anything.

Every call names the project it acts on - there is no ambient "current project" - and access is re-checked against your real Uptimely membership at execute time, so one connection cannot reach a project you are not a member of. An unknown project id and a foreign one return the same refusal, so ids cannot be probed. You connect by signing in to your existing Uptimely account over OAuth 2.1 with PKCE, granting a subset of thirteen granular product scopes at the consent screen; a tool whose scope you did not grant never appears in the tool list and is refused if it is called anyway. Project API keys work for non-interactive clients, and connected assistants can be reviewed and revoked from Connected AI apps in your user settings. The connector processes no payments, generates no images, video or audio, and serves no ads.

20 tools reported — Live, but requires authorization before it will list tools. Publisher lists 20 tools.

For agents

This page has a Markdown and a JSON twin. Pod is also queryable over MCP at https://api.askpod.ai/mcp/read.