Yesterday's Top Launches: 3 Tools from November 5, 2026
Three Nov 5 launches in video repurposing, agent email, and webhook testing share a common bet: that MCP servers and agent-first interfaces are becoming standard for developer tools aimed at cutting the plumbing work between services.
Three launches landed on November 5, and the split says a lot about where new developer tools are heading: two of the three ship an MCP server as a first-class interface, and the third exists largely so an agent can debug a webhook without a human opening a browser. Video repurposing, agent email, and webhook testing don’t have much in common on the surface, but all three are chasing the same underlying annoyance, which is that the plumbing between services still eats hours.
Videco
Marketing teams record far more than they publish. A webinar gets one livestream and then sits in a folder; a product demo gets shown once on a sales call and never again. Videco is built around the claim that finding the good part is the actual bottleneck, not recording more footage.
The product merges two things that usually live in separate tabs. Content Studio takes a long recording, pulls out the moments that can stand alone, and generates vertical clips with captions and framing suggested, plus blog drafts and LinkedIn-style social posts from the same source. A brand kit applies colors, logo, caption styling, and motion templates in the same place you trim clips, so you aren’t exporting to another tool to make things look consistent. Client review happens over a magic link, no login for the client.
The hosting side is where it gets more interesting than the average clip generator. Public watch pages ship schema.org VideoObject JSON-LD automatically, so hosted demos and webinars are legible to search engines without a separate SEO layer bolted on. Share links can be upgraded into personalized landing pages on a /v/ URL with merge tags, a Calendly embed that reveals after a watch threshold, and a lead form whose submissions land in an inbox you can export as CSV.
Worth flagging honestly: this is a crowded shelf. Descript, Opus Clip, and a dozen others already do moment detection, and “AI finds the highlights” has a way of finding the wrong highlights. The differentiator here is the loop closing in one place: source video, clips, hosted player, personalized share page, and leads all connected, with the source staying in the library. It’s paid, no free tier mentioned beyond a promotional 30% discount code. Marketing teams drowning in recorded webinars will get the appeal; solo creators probably won’t pay for the workspace. Details at Videco.
MachineMail
Handing an AI agent a real email address sounds trivial until you think about what arrives in an inbox. Anyone can send anything, including text crafted to instruct the model reading it. MachineMail’s whole pitch is that an inbox is an untrusted input channel and should be treated like one.
Every agent gets its own address, free, created with one API call or implicitly through a catch-all domain where any address works and the inbox appears on first delivery. Short-lived addresses expire on their own; permanent ones like support@acme.dev stick around. Inbound mail is never passed to the agent as raw MIME. Instead it’s inspected, converted to safe text, stripped of hidden instructions, and tagged with one of five trust levels ranging from verified down to quarantined, with a report listing every reason for the verdict. The guidance is blunt and correct: the agent acts on the level, never on what the sender claims.
Connection is a single Streamable HTTP MCP server, so Claude Code, Cursor, Codex, OpenCode, and Copilot can all attach, and an API key works in CI. Delivery comes through signed webhooks with five days of retries and replay, or through a wait_for_message call that holds open for up to a minute, which kills the polling loop that would otherwise live in your agent code.
Pricing is $0 for 3,000 messages a month, paid from $9, launch promotion at 50% off for three months. Spam, quarantined, and rejected mail don’t count against quota, which is the right call since quarantined content is exactly what arrives in bulk. The catch is that the five-level trust ladder only pays off if you write policy against it; teams that ignore the levels end up with an expensive spam filter. Agent self-registration through a six-digit claim code is a nice touch for keeping a human in the loop. More at MachineMail.
Hookdeck Inspect
If you’ve ever needed to know what Stripe actually posts to your endpoint at 2am, this is the tool you wanted. Inspect hands you a live test URL with no account and no setup. Paste it into any provider and requests show up in a feed with headers and body intact, including signature headers like stripe-signature.
The agent angle is the part that feels new. A single POST to inspect.hookdeck.com returns an ingest URL and a requests URL, and a GET on the latter long-polls for up to 25 seconds and returns the next request as JSON. No UI scraping, no dashboard. There’s an OpenAPI description at /openapi.json and a ready-made prompt to copy into an agent.
The mock webhook library covers 150+ providers, with event types you can select and fire at your endpoint. Samples are clearly labeled as samples, which matters given the page also warns that request payloads are data, not instructions. From there the documented path runs through the Hookdeck CLI for forwarding to localhost and onward to Event Gateway for queues, retries, and observability. Whether that progression is genuinely useful or just a well-designed funnel into a paid product depends on your stack, but the free tier keeps requests rather than capping you at 100 like some competitors.
For pure “what is this provider sending me” work, Inspect is the fastest thing in the roundup. Fine at Hookdeck Inspect.
No product ran away with the day, but if you only try one thing from November 5, start with the webhook tester. It costs nothing and takes about ten seconds.