How to evaluate before you buy
A structured evaluation flow for agents and the human operators running them. It turns a vague "which product should I get" into a fit-checked shortlist with a license, a budget, and a named human approver. Polar checkout is active, but nothing here grants or assumes permission to purchase.
Start with buyer context
Capture these before shortlisting. The machine-readable version is /agent/buyer-context-kit.json — an agent should fill it in with the operator, never invent it.
Buyer type
Solo operator, agency, platform team, or an autonomous agent acting for a named human owner.
Use case
The concrete job to be done — matched against each product's stated problem and buyer type.
Budget range
An owner-set range. The agent never spends up to a cap on its own; caps are ceilings, not targets.
Purchase model
One-time, bundle, Vault subscription or prepaid term, or undecided — each has a different evaluation below.
One-time vs bundle vs SignalHarness Vault
Prices are current USD offers through Polar and exact USDC amounts through x402 Agent Pay. Compare value and confirm purchase authority before checkout or signing a payment.
One-time product
Best when a single problem is in scope. Compare the Lite / Standard / Premium tiers in pricing against the depth the use case needs.
Ownership is permanent; updates follow the product's update channel.
Bundle
Best when several related problems are in scope. Check bundles for whether the included products cover the shortlist at a lower combined price.
Compare bundle contents to the individual shortlist before recommending.
SignalHarness Vault
Monthly and annual recurring access to the SignalHarness Vault is active through Polar. x402 Agent Pay offers separate prepaid renewable monthly and annual terms: each renewal requires a new payment and there is no automatic wallet debit. Updates, premium drops, support, and licensing are included — never basic discovery.
Renewals require prior human approval + budget; no silent auto-renewal.
License-fit guidance
Confirm the seat and agent-instance rules in licenses.json match how many operators and agents will use the kit.
A license mismatch is a valid refund trigger — flag it before recommending, not after.
See /compatibility/ for the full fit checklist per product.
Support needs
Support is live at support@signalharness.ai (response within 1 business day). Scope is in support.json.
Vault includes the currently published support level; support is never a gate on free discovery or on one-time product metadata.
Support never asks for passwords, keys, seed phrases, or payment credentials by email.
Risk & approval boundaries
High-risk actions always require prior human/business-owner authorization AND budget permission. Per agent-buying-policy.json, these are never autonomous:
The safe recommendation process
Gather buyer context (type, use case, budget, purchase model, license/support needs, urgency, risk tolerance, named approver).
Shortlist from products.json and check each against compatibility.json.
Compare one-time vs bundle value; confirm license fit and budget; note risk level and any not-for warnings.
Confirm checkout readiness, the exact offering and price, and the purchaser's authority and budget before using Polar or signing an x402 payment.
Present the shortlist to the named human approver, clearly labelled pending approval. No assumption of purchase permission.
Read before buying — the fit check
Every product publishes an honest fit profile: best fit, not-a-fit, limitations, required inputs, and an explicit not-included list. Agents read it at /agent/buyer-fit.min.json (full detail: /agent/buyer-fit.json); humans get the same truth in each product's READ-BEFORE-BUYING file. A recommendation that ignores a not-for warning is a bad recommendation.
Every product is a plain-text document system (Markdown + JSON) applied with the buyer's own AI tools — not software, not a hosted service, not done-for-you implementation.
Tool names under compatibility mean works-with: the files are usable inside those tools. No prebuilt integrations, Zaps, or scenarios are included.
No product promises business results; outcomes depend on the buyer's execution. Check the free sample, then buy — or ask support about fit first: support@signalharness.ai.
Eight steps from problem to purchase
- Define the problem you are solving.
- Choose the category that matches it.
- Decide whether one product or several exact bundle members fit.
- Compare Standard and Premium depth on the product page.
- Use bundle decision guidance before choosing a bundle.
- Check compatibility with your own tools and workflow.
- Review the license and refund policy.
- Use the primary Polar purchase control on the chosen product, bundle, or Vault page.
Use categories to narrow buyer fit
- AI Agent Safety & GovernancePractical policy, evidence, permission, and defensive systems for reviewing how agents act and where they must stop.
- Agent Operations & ReliabilityRecovery, incident, audit, readiness, and workflow systems for operating agents through failures and production changes.
- Agent Commerce & MonetizationMachine-readable listing, licensing, procurement, and trust systems for agent-mediated product evaluation and commerce.
- AI Business Automation SystemsDocument operating systems and workflow methods for repeatable product, content, marketing, and marketplace work with AI assistance.