B2B Sales Outreach Skill
Overview
Your submission is a single SKILL.md file that teaches a fixed base agent to craft cold B2B sales outreach sequences — structured, multi-touch email campaigns built from an unstructured prospect brief.
The core objective is straightforward: given a prospect brief, the agent must produce a correctly formatted numbered table (Touch # | Send Day | Hook Type | Subject Line | Body | Personalization Notes) followed by a concise strategic summary that explains the trigger anchor, sequence length rationale, and any enrichment flags.
Strong skills teach the agent a reasoning procedure it can apply to any prospect brief it has never encountered. That means: classifying the prospect's seniority tier and vertical to set tone and length; ranking available trigger signals to anchor the first touch; choosing an appropriate sequence length; assigning distinct, non-repeating hook types across touches; and pacing CTAs correctly (interest-checks early, calendar asks from Touch 3 onward). The skill also teaches the agent when to stop and ask for missing information rather than fabricate specifics, and when a request falls entirely outside the scope of cold outreach.
Weak skills hand the agent templates or canned sequences that only work when surface details match what the skill author imagined. They break immediately when the vertical shifts, the seniority level changes, or no clean trigger signal exists. A skill that drills the agent on memorised sequences for a handful of archetypes will fail the moment a brief lands with an unfamiliar combination of signals.
A skill built to memorise specific scenarios will not hold up as challenges evolve.
You submit a SKILL.md file to the challenge. The validator attaches it to our canonical base agent, runs it against 10 sampled scenarios, and returns a scored breakdown. Top submissions earn Alpha emissions daily and on displacement.
We are looking for skills that demonstrate good design and help the base agent behave like a capable practitioner in this domain.
Requirements (10 Tests)
| Category | Overview | Test cases |
|---|---|---|
| Type A - Multi-Signal Sequence Coherence | This type tests whether the agent can build a sequence that holds together as a whole, not just a collection of individually acceptable emails. Scenarios require the agent to vary opening structures and email lengths, compound a single thesis across touches without redundancy, and ensure every touch is independently readable by a prospect who missed earlier ones. The skill must teach the agent to treat coherence as a structural discipline, not an afterthought. | |
| Type B — Trigger Signal Judgment | This type tests whether the agent can identify, rank, and act on trigger signals — or reason clearly in their absence. Scenarios range from briefs rich with fresh, specific signals to briefs where no meaningful trigger exists. The skill must teach the agent to assess signals by recency, specificity, and product relevance, select the strongest anchor for Touch 1, and apply appropriate send-day compression or expansion based on trigger urgency. | |
| Type C — Scope and Enrichment Judgment | This type tests whether the agent correctly recognises the boundaries of cold outreach and handles information gaps responsibly. Scenarios include requests that fall outside scope (warm intros, inbound replies, renewals) and briefs that omit information critical to personalisation. The skill must teach the agent to redirect out-of-scope requests cleanly and to surface specific missing fields with a reasoned enrichment request rather than fabricating plausible-sounding details. | |
| Type D — Hook Architecture and Thesis Development | This type tests whether the agent can architect a sequence with a distinct, non-repeating hook type on every touch and a thesis that develops logically from trigger-based opening through insight, social proof, direct ask, and breakup. Scenarios vary the number of required touches (3, 4, or 5) and the richness of available context. The skill must teach the agent to assign hook types as deliberate structural choices and to develop the product-to-prospect argument progressively across the sequence. |
Surface details — prospect names, company names, AUM figures, headcounts, funding amounts, trigger events, and product descriptions — rotate across variants while the underlying capability being tested remains constant.
Datasets
Available Datasets
Evaluation Criteria
Each scenario is graded by an LLM judge. Final scores combine two graders and a set of deterministic penalties.
| Criteria | Weight | |
|---|---|---|
Skill use An LLM judge evaluates whether the agent's output actually followed the procedures your skill teaches, rather than freelancing or pretending to use the skill. | 30 % | |
Scenario quality An LLM judge evaluates how well each output meets the per-type quality bar, independent of whether your skill was followed. A reference good output anchors the judge's sense of the quality bar. | 30 % | |
Rubric compliance An LLM judge scores each output against a structured rubric for this challenge's communication artifacts — audience calibration, structure, and actionability. The rubric itself is not published: write for the reader each artifact serves, not for a checklist. | 30 % | |
Output Novelty Outputs are compared across scenarios. Near-duplicate, template-style responses that reuse the same content regardless of scenario specifics are penalized. | 10 % |
Scenarios are scored individually; deterministic penalties apply to outputs that damage quality in ways the rubric treats as clear failures (for example, fabricating specifics the brief did not provide, repeating hook types, or applying the wrong CTA type for a given touch number). Individual scenario scores aggregate into per-capability-type averages, and those averages combine into the final challenge score shown on the leaderboard.
Submission
How to submit
Go to your sundae_bar account settings under wallet & transactions and verify you have your wallet id verified. If your accounts hotkey is registered to SN121 you will see it below with your miner UID. Otherwise you can use our prompts to register or register on the command line.
Make sure you review all requirements before creating your skill.md file
On this page there is a submission button. If active and a registered miner to SN121 you can submit your contribution.
Note: we have daily user limits and total rolling limits.
First your submission will go through anti-gaming gates. If passed the validator will run the evaluation suite on your submission to return your score.
Constraints
Constraints details coming soon...
Rewards
Emissions accrue on-chain to whichever submission currently holds the top of the leaderboard. There is no fixed per-submission payout: the current leader earns for as long as it stays first.
A strong skill submitted early starts accruing immediately and keeps earning until another submission displaces it. Once you are displaced, accrual stops until you reclaim the top.
Read this article for more details.