Ingest your product. Find your ICPs. Map the competition. Ship the plan. Validate demand. Watch what changes. The full GTM motion — end to end.
Ingests every signal — site, docs, market chatter, your own input — and builds a product context that every downstream skill uses. The level of understanding a great fractional CMO would have after a deep-dive week. In minutes.
Identifies the buyers most likely to pay. Writes the positioning. Maps the competition. Builds a channel plan with specific actions, sequencing, and metrics — not generic "do content marketing" advice.
Watches your competitors and the incumbents in your buyer's stack. When something material changes, it tells you what to update in your positioning, plan, and pricing — and shows the cascade.
Listed three angle variants on Facebook Marketplace at $29/mo and watched the engagement for 5 days.
Verdict: real demand. The "guilt" angle clearly outperforms the "automation" angle. Build at $29 entry price; pivot copy to the founder-weekly framing.
Ingested the site, docs, your two LinkedIn posts, and the relevant subreddits. Built the product context.
What Loop is: an LLM that reads a founder's GitHub, Linear, and calendar each week, then drafts a newsletter in their voice.
What's distinctive: work-in-real-time drafting (not retro summaries), founder-voice match, sub-10-minute weekly edit loop.
What's underdeveloped in your messaging: the "voice match" claim is the strongest moat — your site buries it on page 3.
Researched four ICP candidates. Two have clear pull, two are distractions.
Wrote up a positioning doc with a messaging matrix and a pricing recommendation ($29 / $99).
Six players matter. Two are real threats, four are noise.
Mapped the full landscape — pricing, community sentiment, and a feature-watch list for Beehiiv.
Four channels in priority order. Two are owned (run them now), two are leveraged (start week 3).
Built the full plan — success metrics, weekly cadence, and the "decline to run" reasoning per channel.
Three changes since last week's scan. Two are signal, one is noise.
Compiled a delta report with screenshots and the trigger to run gtm-reassess on Beehiiv's launch.
This is material. Three documents need updates — here's the cascade.
Prepared an impact assessment with section-level diffs. Approve and I'll apply the changes.
The seven skills above are one pipeline on Curunir — a deliberately minimal agentic harness. The same engine runs every step, and nothing more. /launch/ is one pipeline; the same harness can run research, scheduling, or anything else you wire into it.
It runs on a small device in your home or office. The language models live in the cloud — but your roadmap, your customer list, and your competitive intel don't. They stay on the device.
Reachable from anywhere — web, CLI, email — through a thin portal that brokers the conversation without copying the data.