Meeting Follow-Up Communication Skill
Overview
Given a structured meeting recap — including decisions, action items with owners and due dates, parking-lot items, and attendee lists — the agent must produce a single ready-to-send follow-up email: correctly tiered subject line, calibrated salutation, and a structured body that puts every recipient's relevant content exactly where they expect to find it.
Your submission is a single SKILL.md file that teaches a fixed base agent how to perform meeting follow-up communication. The skill must convey the full reasoning procedure: how to read a recap, classify every recipient, select the right urgency tier, sequence and format action items, handle absentees, and close with a concrete next step — all before writing a single word of the email draft.
Strong skills teach the agent to recognise the specific situation in each recap (who attended, who owns what, what is missing, how urgent the actions are) and apply a defined procedure to produce a correct, send-ready output. Weak skills hand the agent a fill-in-the-blank template; those break the moment a recap has an unusual attendee mix, a missing due date, or a sender who is also an action-item owner.
The skill must also teach the agent when not to draft: how to flag incomplete data back to the requester before proceeding, how to refuse to fabricate details not present in the recap, and how to redirect out-of-scope requests rather than attempting them.
A skill built to memorise specific scenarios will not hold up as challenges evolve.
You submit a SKILL.md file. A fixed base agent loads it and answers all 10 scenarios with it, then returns a scored breakdown. The submission that holds the top of the leaderboard accrues on-chain emissions for as long as it stays first.
We are looking for skills that demonstrate good design and help the agent perform like a capable practitioner in this domain.
Requirements (10 Tests)
Your skill is tested across 5 capability types, with 10 scenarios in total. Every submission is evaluated against all 10 scenarios — there is no sampling, so consistent quality across every capability type counts.
| Category | Overview | Test cases |
|---|---|---|
| Type A — Action Item Sequencing & Rendering | Tests whether the agent can correctly group action items by owner, lead each item with its due date, order items from earliest to latest due date, and include acceptance criteria only when the recap explicitly provides them. The skill must teach the full rendering procedure so the agent applies it consistently regardless of how many owners or items appear in the recap. | 1–3 |
| Type B — Minimal-Content Recap Structuring | Tests whether the agent can produce a correctly structured follow-up email when the recap is sparse — few decisions, few or no action items, or a thin parking lot. The skill must teach the agent to select the appropriate subject-line urgency tier and assemble the email body sections in the defined order even when several sections have little or no content to fill. | 1–3 |
| Type C — Mixed Recipient Classification | Tests whether the agent correctly classifies every recipient in a recap that contains a mix of attendees, absentees, action-item owners, and a sender who may themselves own an action item. The skill must teach the classification procedure — including the sender-as-owner edge case — so the agent routes content and salutation correctly for any combination of recipient classes. | 1–3 |
| Type D — Multiple Actions, Single Owner | Tests whether the agent correctly consolidates all action items belonging to one owner into a single grouped block, ordered by due date, without splitting or duplicating them across the email. The skill must teach the grouping and ordering logic so it holds even when a single owner has many items with varying due dates and mixed presence or absence of acceptance criteria. | 1–3 |
| Type E — Pre-Draft Flagging & Scope Boundaries | Tests whether the agent correctly halts and flags incomplete data (missing owners or due dates) before drafting, refuses to fabricate details absent from the recap, drafts accurate absentee catch-up paragraphs that name the absentee and surface personal asks explicitly, and redirects out-of-scope requests. The skill must teach the agent to recognise each of these situations and respond with the appropriate action rather than proceeding to draft. | 1–3 |
Surface details — names, dates, meeting titles, action-item specifics, and attendee compositions — rotate across scenario variants, while the underlying capability being tested remains the same.
Datasets
Available Datasets
When you submit, we run your SKILL.md through a standardised base agent provided by sundae_bar. The base agent is the same for every submission and is not supplied by you. It receives the challenge's test scenarios one at a time, calls the load_skill tool to read your skill, and follows it to answer each scenario.
Same base agent, same tests, only the submitted skill changes. Scores therefore reflect the quality of your skill, not agent configuration: the agent's answers are only as good as the procedure your skill teaches.
Graders then score each answer for output quality, correct skill use, and gaming behavior, and the per-scenario scores are averaged into your challenge score.
Important: the base-agent harness is fixed at evaluation time and its available tools are load_skill. You submit only a SKILL.md; you cannot supply your own agent, add tools, or alter the base agent's configuration.
Tip: write your skill as though you are onboarding a capable new hire who needs a procedure, not scripting a chatbot.
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. | 50 % | |
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. | 50 % |
Scenarios are scored based on the quality of the agent's output across multiple dimensions; deterministic penalties apply for output moves that meaningfully damage quality — such as fabricating content, misclassifying recipients, or producing a malformed structure — and these penalties are applied consistently regardless of how well other aspects of the response perform. Individual scenario scores aggregate to capability-type averages, which in turn roll up to the overall challenge score reported on the leaderboard.
Submission
How to submit
Create an account and add your Bittensor wallet ID to your profile to earn alpha rewards in the challenge.
Review the reference skills (good, mediocre, gaming) to understand the quality bar and calibration points.
Draft your SKILL.md. Focus on teaching procedure rather than providing templates.
Validate locally using the published dataset variants to sanity-check output quality.
Submit through the sundae_bar challenge portal by uploading your SKILL.md file.
Review your scored breakdown and resubmit.
Constraints
Submission format: a single SKILL.md file with YAML frontmatter (name, description) and Markdown body.
Base agent: fixed. Provided by sundae_bar.
Runtime: a fixed base-agent harness loads your SKILL.md and answers all 10 scenarios per submission with it.
Tests per submission: 10 (the full scenario set — no sampling).
Submission limits: 3 submissions per day. A rolling challenge limit also applies.
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.
Time at the top takes all. 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.
Rankings shift as others iterate. Get to the top early, then keep refining your skill to defend the position.
Your SKILL.md is scanned at submission time by a skill integrity gate. The gate checks that your skill teaches procedure rather than encoding scenario-specific content from our evaluation pool.
A skill that fails the integrity gate scores 0 for the submission. The gate runs deterministically and reports its finding as part of your results. This is how the evaluation framework stays credible: a high score should reflect a skill that genuinely teaches procedure, not one that memorised our test set.