Contenu du pack

Un fichier SKILL.md (le guide pratique) + un modèle de départ. Aperçu :

Agent-Economy Stack Mapper

When to use

• Customers' agents are already interacting with your business (or will be next quarter) and you have no map of where.
• You're deciding where to invest in agent-readiness: identity, payments, APIs, support.
• Leadership asks "what's our agent strategy" and you want an answer grounded in your actual touchpoints.
• You're a vendor selling into the agent economy and need to speak your buyer's language.

Workflow

1. List the touchpoints (input: your business → output: touchpoint inventory)
   Walk the customer journey and mark everywhere an agent could act: discovery (agents comparing your product), purchase (agents buying), support (agents filing tickets), procurement (agents negotiating renewals), identity (agents proving humanity), and post-purchase (agents managing subscriptions, returns). Include the touchpoints you don't control — review sites, marketplaces, social — where agents already summarize you.
2. Classify each touchpoint (input: inventory → output: classified map)
   Tag each: who sends the agent (your customer? a third party? you?), what it can do there today (read-only? transact? modify?), and what's at stake (money, data, reputation, legal). The high-stakes transacting touchpoints are your investment priority; the read-only ones are your monitoring priority.
3. Score agent-readiness (input: classified map → output: readiness scores)
   Per touchpoint, score 1–5 on: machine-readable (can an agent parse your offering without a human?), identity-aware (can you tell a verified-human-backed agent from a bot?), transactable (can an agent complete the action with proper authorization?), and observable (do you log agent interactions separately from human ones?). Anything scoring under 3 on identity-aware is where fraud will arrive first.
4. Match trust infrastructure (input: scores → output: infrastructure plan)
   For each priority touchpoint, name the infrastructure: delegated identity proofs (so agents prove a human principal), scoped payment credentials (so agents spend within bounds), structured APIs or agent protocols (so agents don't screen-scrape), and audit trails (so every agent action is attributable). Mark build vs buy vs wait per item — not everything needs building this quarter.
5. Write the agent policy (input: plan → output: one-page policy)
   Your public stance, in plain language: which agents you welcome, what they may do, how they identify themselves, what happens to unverified bots (rate-limited? blocked? served read-only?), and where to get credentials. Publish it — the agent economy rewards the businesses agents can understand.
6. Set the review cadence (input: everything → output: quarterly review)
   Agent capabilities move quarterly; your map must too. Each quarter: re-score readiness, check the fraud signals (bot traffic share, disputed agent orders, support tickets filed by agents), and move one touchpoint up the maturity ladder. The map is a living doc, not a deck.

Aperçu du modèle

Agent-Economy Stack Worksheet

Map where agents touch your business — then decide what each touchpoint needs.

1. Touchpoint inventory

2. Classify

Investment priority = high-stakes + transacting. Monitoring priority = read-only.

3. Readiness scores (1–5)

Under 3 on identity-aware = where fraud arrives first.

4. Infrastructure plan

5. One-page agent policy (publish it)

• Which agents you welcome:
• What they may do:
• How they identify themselves:
• Unverified bots get (rate-limited / blocked / read-only):
• Where to get credentials:

6. Quarterly review

• Re-score readiness: date
• Fraud signals (bot traffic share, disputed agent orders, agent-filed tickets):

Le pack de compétence lui-même (SKILL.md + modèle) est rédigé en anglais.

Acheter le pack complet — $4.99 Pro — les 511 compétences