@clossys/customer

First-person synthetic inhabit of the named Strategist audience — the person this was made for, on speed dial. Not a hired QA reviewer, auditor, contractor scorecard, or third-person market-research memo.

npm install @clossys/customer

This package is published to the public npm registry, https://registry.npmjs.org. Installing it needs no authentication: no npm token, no .npmrc registry override, and no GitHub credential of any kind.

Job question

If I am the person this was made for, do I keep this?

The package name customer means the person the product is for — not a paying-buyer-only legal status and not Bouncer's authenticated user. This package is not a reviewer.

That job is the seal gate. The same inhabit also answers on-demand as that person: a synthetic power user you can ask about any topic, comparison from their actual consideration set, what it would take to start or to refer, whether it is worth what it costs them, and what would make them leave.

Session intents

customer-check validates inhabit form for every session. Only keep counts toward the charter metric.

| intent | What the person speaks | Charter metric | | --- | --- | --- | | keep | Fresh or returning look; visual and verbal; one keep/fail | yes | | feedback | Lived functional stumbles, experience, missed expectations — any topic | no | | compare | Alternatives I actually use, a peer uses, or I considered, from my day | no | | refer | Whether I would tell a peer, the words I would use, what it would take | no | | churn | The warning, the moment I would leave, where I would go, what would keep me | no | | adopt | Whether I would start, what stops me, the first real job I would give it | no | | worth | Whether this is worth my time, money, or attention, and the threshold | no |

Every session binds speaker: "customer", inhabitedAs: "target-audience", and the named Audience. Missing intent is treated as keep. Unknown intent cannot run. Speed-dial testimony is the same person, not a second role: not QA, not Strategist intel, not Influencer yield.

Feedback may be purely functional, purely experiential, or both. At least one lived channel (functional, experience, or expectations) is required. Compare with an empty consideration set is a finding.

Customer keep rate

Independent consumer evidence shows the position's owned metric meets its setpoint over the declared review cadence. The owned metric is customer keep rate, computed by assessCustomerKeepRate(). An empty evaluated set is indeterminate, never a rate of 1. Feedback, compare, refer, churn, adopt, and worth testimony never enter this rate. customer-check proves the session was conducted as the customer; it never certifies a five-star quality score. This package does not measure consumer evidence and does not close the loop. A green run of this package's tests is not a close.

import {
  assessCustomerKeepRate,
  checkInhabitForm,
  checkKeepForm,
  parseAudience,
  parseInhabitRecord,
  parseKeepRecord,
} from "@clossys/customer";

const report = assessCustomerKeepRate(input);

customer-rate-check assessment.json

The command prints JSON and exits 0 for satisfied, 1 for violated, and 2 for indeterminate, unreadable, or invalid input.

This package declares that command as its first-day assessment surface in its own manifest:

"foundry": { "assessment": { "bin": "customer-rate-check", "invocation": "single-json-input" } }

Onboarding discovers that declaration from the installed manifest and never infers a surface. customer-check remains the two-argument CLI and is not the assessment surface.

The same foundry block also declares, for discovery and never for inference:

• outputs — the files this role owns in a consumer repository, one per session intent: clossys/customer/keep.json, feedback.json, compare.json, refer.json, churn.json, adopt.json, and worth.json. customer-check reads one such record at a time; nothing in this package writes them.
• feeds — the one artifact another role consumes: the keep-verdict at clossys/customer/keep.json, which Publisher waits for before it seals.
• fit — fit-signals.json, shipped in this package: the evidence that makes this role applicable at all (a named audience to speak as, and an audience-facing candidate to keep or fail).

Inhabit form gate

customer-check record.json audience.json

Exit codes: 0 clean form, 1 inhabit or consistency findings, 2 could not run (including unreadable files, unknown intent, or indeterminate shape).

Audience is a JSON seam { id, name, description, painPoints? } with no runtime dependency on Strategist. parseAudience, parseKeepRecord, parseInhabitRecord, checkKeepForm, and checkInhabitForm are the inhabit-form surface (checkKeepForm is an alias of checkInhabitForm). Exported session types are KeepRecord, FeedbackRecord, CompareRecord, ReferRecord, ChurnRecord, AdoptRecord, and WorthRecord. A keep verdict with any impression answered no is a finding. Compare with an empty consideration set is a finding. Speaker designer, qa, or reviewer is a finding.

Exported types

Session records:

| Type | What it is | | --- | --- | | Audience | The JSON seam { id, name, description, painPoints? } a session binds to. | | InhabitEnvelope | Fields every session shares: speaker: "customer", inhabitedAs: "target-audience", audienceId, persona.name, stance, topic, familiarity, and intent. | | InhabitRecord | The union of every parsed session: KeepRecord, FeedbackRecord, CompareRecord, ReferRecord, ChurnRecord, AdoptRecord, or WorthRecord. | | InhabitIntent | "keep", "feedback", "compare", "refer", "churn", "adopt", or "worth". | | Familiarity | "fresh" or "returning". Required for every intent except keep, where it defaults to "fresh". | | YesNo | "yes" or "no", the only accepted answer to every yes/no field. | | KeepVerdict | "keep" or "fail", the one verdict a KeepRecord carries. | | KeepImpressions | A keep's firstSeconds sentence plus its four YesNo answers: isThisForMe, doIBelieve, wouldIStay, and wouldITellAPeer. | | KeepChannelImpression | { impression }, the shape of a keep's visual and verbal channels. | | LivedFunctional | One feedback item: what happened and what I expected. | | LivedExpectation | One feedback item: what I assumed and what actually happened. | | KnownAlternative | One entry in a compare's consideration set: name, relationship, and whyItMatters. | | AlternativeRelationship | "i-use-this", "a-peer-uses-this", or "i-considered-this". |

Inhabit-form results, from checkInhabitForm and checkKeepForm:

| Type | What it is | | --- | --- | | KeepFormReport | { state, findings }. | | KeepFormState | "satisfied", "violated", or "indeterminate", matching customer-check exit codes 0, 1, and 2. | | KeepFormFinding | { rule, severity: "error", message, path? }. |

Keep-rate results, from assessCustomerKeepRate:

| Type | What it is | | --- | --- | | CustomerKeepRateAssessment | { metric: "customer keep rate", state, rate, evaluatedCandidates, keptCandidates, findings, proposedPositions }. rate is null whenever no declared candidate has a counting independent observation, and the state is then indeterminate. | | CustomerKeepRateState | "satisfied", "violated", or "indeterminate", matching customer-rate-check exit codes 0, 1, and 2. | | CustomerKeepRateFinding | { rule, severity: "error", message, path? }. |

Metric definition

customer keep rate
= independently observed first-person keeps
  / all candidates independently inhabited before seal or land

The unit is a ratio and the desired direction is up.

Changelog

Release notes for every version are in the changelog, kept in the public repository rather than in the installed package.