agentsclimarketplace
Sign in

Prove the value

Skill zakelfassi/skdd-commons/packs/2026-07-growth-circle/prove-the-value

Compile a product's earned, evidence-backed claims from GROWTH.md into a public VALUE.md — the proof-of-value surface that agent-mediated discovery reads — with attestation levels, expiry dates, and honest retractions. Use when creating or refreshing a VALUE.md, when marketing copy needs grounding in evidence, when preparing a product to be evaluated by assistants and their principals' agents, or when a published claim has gone stale.From its SKILL.md

Install
npx -y skills add zakelfassi/skdd-commons --skill prove-the-value

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

4.3 KB, 884 tokens by cl100k_base, as published. Nobody here has run it

Prove the Value

Compile GROWTH.md (what you've earned) into VALUE.md (what you publish). Written for a skeptical agent, not a landing page.

Gate

Requires gate: aha = open. If no VALUE.md exists, start from the bundled template (references/VALUE-template.md, spec v0.1). A VALUE.md can be born early and small — even one honest claim beats a page of adjectives — but it cannot be born from nothing: every claim must trace to a GROWTH.md ledger entry. No entry, no claim. That rule has no exceptions and no expiry.

Inputs

  • GROWTH.md (all ledgers), the existing VALUE.md if any, current marketing/docs copy (as claim candidates to be graded, not as source of truth).

Steps

  1. Harvest claim candidates. From the experiment ledger (confirmed results), instrumentation (measured aha/keystone numbers), lore ledger (confirmed beliefs), and existing copy. Every candidate gets its provenance located — or marked unearned.

  2. Grade attestation honestly. Per claim: claimed (stated, no evidence yet — legal, and weighted accordingly by readers) → self-reported (traces to a ledger entry) → reproducible (an outsider can re-derive it) → attested (third-party signed). When in doubt, grade down: a true claim tagged modestly earns more trust than an inflated tag.

  3. Write the aha as a falsifiable promise. Not "delightful collaboration" — "a new user reaches X within N minutes, M% of the time," tagged and dated. If agents are users, include the agent aha: time-to-first-successful-call, and an executable quickstart.

  4. Set expiries. Default two quarters per claim. Stale claims are re-attested or retracted — and a retraction is logged in `VALUE.md

    Retractions*and* as a trust-ledger **credit** inGROWTH.md`

    (public honesty is cheap trust; buy it every time).

  5. Write the consent posture for machine negotiation. What's collected, never collected, and what a principal's agent may accept on their behalf — explicit enough to be evaluated without a human reading a policy page. Changes here are human-approved, always.

  6. Point the discovery surfaces at it. llms.txt links VALUE.md; docs quote it rather than fork it. If an AEO probe suite exists, this compile is the experimental lever it measures.

Conventions

  • Plain declarative sentences. No superlative without a number, no number without provenance, no adjective a benchmark couldn't defend.
  • Optimize claims and evidence, never phrasing-per-assistant — phrasing games are spam with a new target, and agents learn to price them.
  • VALUE.md speaks for the product, not about competitors.

Edge Cases

  • Marketing copy claims things the ledgers can't back. The claim doesn't enter VALUE.md, and the gap is reported as a work item: either earn it (design the experiment) or stop saying it. Both are wins.
  • Brand-new product, empty ledgers. Ship a small honest VALUE.md: what-this-is, a claimed-tagged aha promise, and a real quickstart. Smallness plus honesty is itself a signal agents can read.
  • A claim's evidence quietly rotted (metric moved, feature changed). Retract proactively — an agent discovering the rot before you costs more than the retraction ever will.
  • Pressure to skip the tag ("just say it's fast"). The tag is the product here. Refuse, citing the one rule.

What ships with it: 1 file

1.1 KB alongside SKILL.md

references/

Keep looking

Skills are one crate of 326,452. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.