New: two skills for email hero design and IP warming, plus clearer signals for when Orbit is doing the work
Two new skills join the library. email-header-design runs a multi-agent loop that researches a brand from its live site, generates a hero image, and has a panel of design reviewers critique and re-prompt it until the design clears a quality bar: for the banner at the top of a marketing email. ip-warming does the maths for micro IP warming, spreading a one-time full-base send across several days so daily volume never spikes above what the IP's recent history earned, then shows you how to build it in Braze with Audience Paths, Random Bucket Numbers, and Delays, including the live Braze MCP method for pulling your trailing-30-day send volume when Braze exposes no single rollup for it. Alongside the skills, heavy Orbit tools are now easier to spot: in Claude Desktop they carry a branded chip title like "Orbit · Braze Sync", and their replies open with a concrete outcome line naming what Orbit did and the real result numbers, not just the closing signature.
What shipped
•New skill: email-header-design. It treats hero design as a process, not a one-shot generation: an agent pulls the brand's palette, type, and style from the live website, then several rounds of AI image generation are each critiqued by a panel of reviewers (brand fidelity, mobile legibility, composition) and the feedback is synthesised into the next prompt, looping until a quality score is hit. It produces a finished hero plus a reviewer scorecard. Because it spawns roughly ten to twenty agents per run, it needs multi-agent orchestration and explicit opt-in, and it's for the top-of-email banner only: not in-body images, full email builds, or anything that should stay vector.
•New skill: ip-warming. The focus is micro IP warming: your IP is already warm, but a one-time send to the full base would spike daily volume far above the norm, and mailbox providers read a sudden 5x jump as a different sender and throttle it. The skill anchors every daily cap to your baseline volume (trailing-30-day sends divided by 30), then spreads the single send across N days with escalating random-bucket splits and delays so no day exceeds what recent history earned. The maths mirrors the IP Warm-Up Planner web app; the skill adds the Braze build via Audience Paths, Random Bucket Numbers, and Delays. For cold-IP or new-domain warming over weeks, it defers to the deliverability-management skill instead.
•ip-warming carries the live Braze read method. Braze exposes no single "total email sends" series, so the skill enumerates and sums: paginate the campaign list, sum the email-channel sent field across each campaign's 30-day data series, then add the canvas sends. It flags the trap that canvas series caps at 14 days, so a true 30-day canvas total needs two rolling windows summed, or a 14-day proxy that's labelled as such, never a silent undercount.
•Behaviour change: heavy Orbit tools now show a branded chip in Claude Desktop. Tools that do real external work read as "Orbit · Braze Sync", "Orbit · Lifecycle Diagram" and the like in the tool UI, matching the "Built with Orbit" reply signature, so it's obvious at a glance when Orbit is doing the work rather than generic Claude. Light passthrough tools keep their plain titles on purpose: branding every tool would dilute the signal.
•Behaviour change: heavy-tool replies now lead with the outcome. On top of the existing closing signature, a heavy tool's reply opens with one concrete line naming what Orbit did and the real result numbers: for example "Published 6 modules to your live Braze workspace via 7 REST calls" rather than a generic "Published Orbit assets". The figures come from the actual tool response; the reply never manufactures numbers the tool didn't return.