reply-team/jason-ai / skills/business/sdr-operations
sdr-operations
Verifiedlexicon/neutral.yaml
View on GitHub# The neutral register of the SDR business operations contract.
#
# The contract owns the CONCEPTS, and the words it will ACCEPT on input. A provider adapter owns
# the words a product DISPLAYS. Aliases in, labels out — this file is the "in" half, and it emits
# nothing. A plan, a report and an operation name are always written in `prefLabel`; the surface a
# user reads may call it whatever that surface calls it, and choosing that word is the adapter's
# business, not this file's.
#
# THE ACCEPTANCE TEST. Delete every dialect pack and the contract still works — only the words a
# user sees change. If something breaks, a concept leaked into a dialect and belongs here.
#
# ---------------------------------------------------------------------------------------------
# SCHEMA — one entry per concept, under `concepts:`
# ---------------------------------------------------------------------------------------------
#
# - id: campaign # the contract's entity, exactly as it appears in an operation name
# prefLabel: campaign # the one word the contract writes. Never an alias, never a plural.
# altLabels: [cadence] # accepted on input, NEVER emitted. Spellings, regional variants,
# # and the words practitioners actually say for this concept.
# hiddenLabels: [cmp] # searchable, never displayed and never written back. Misspellings,
# # abbreviations, plurals and inherited column headers live here, so
# # a lookup succeeds without the word ever reaching a reader.
# definition: > # what the concept IS, in one or two sentences, with no product in it
# never_means: # senses this word carries elsewhere that do not exist here. Optional,
# - context: marketing # and present only where the confusion is real and load-bearing.
# sense: > # the other reading, stated plainly enough to be recognised
# because: > # why it cannot be that here — the structural reason, not a preference
#
# `context` is a key, not prose: it names the register the other sense comes from, so a resolver
# can report "you are using the marketing sense of this word" rather than "unknown". Values are
# free but stable — reuse an existing one before inventing another.
#
# WHAT IS NOT HERE. Labels a product shows. Field names. Anything that varies by installation:
# an account's own vocabulary — stages, dispositions, reply categories, meeting outcomes — is read
# with `vocabulary.list` and is an account fact, not a contract fact. A word nobody actually says
# is not an alias; do not invent input we have never received.
#
# WHERE prefLabel GOVERNS, AND WHERE IT DOES NOT. It governs everything written back: an
# operation name, a plan, a report, a durable artefact, an identifier, a machine-readable field.
# It does not govern explanatory prose, which may reach for an alias where the alias reads better
# — a skill saying "the prospect sees a new From address" is using the accepted input word for
# `contact` in a sentence nobody parses. The line is the sense, not the word: an alias must never
# be used in a sense this register denies (see `never_means`), and it is never the word a plan or
# a report writes.
#
# ONE DOCUMENTED SPELLING SPLIT. Identifiers are US-spelled — `enrollment.pause`, `campaign.enroll`
# — because an operation name is a token and tokens do not get regional variants. Prose in this
# repository is UK-spelled, so it says "enrolment" while naming `enrollment.list` in the same
# sentence. That looks like an inconsistency and is a deliberate one: the identifier is fixed by
# the contract and the surrounding sentence follows the repository's own English. `enrolment` is
# an accepted input alias for exactly this reason.
#
# HOW A WORD IS RESOLVED AT RUNTIME. `term.resolve` answers with the concept, or with `ambiguous`,
# or with `unknown` — never with a guess. This file is what it resolves against. An alias resolving
# to two concepts is a defect here, not an ambiguity to be reported at runtime.
version: 2.0.0
register: neutral
# ---------------------------------------------------------------------------------------------
# RETIRED WORDS — not aliases, and never resolved to a concept.
#
# NOT A SECOND DECLARATION. `operations/families.yaml` declares the retirement and the build reads
# it from there; if the two ever disagree, that file is right. They are repeated here because a
# word lookup starts in the lexicon, and a retired word must come back as *retired* rather than as
# *unknown*: a sentence written today that says `sequence.activate` or `metrics.campaign` is naming
# something that does not exist, and the reader deserves to be told which, instead of having the
# token pass as ordinary prose or as a typo.
#
# They are deliberately NOT listed as altLabels of the concepts that replaced them. An alias
# resolves silently; that is the whole point of an alias, and it is exactly the wrong behaviour
# here. A renamed entity is a defect in the sentence, not a dialect to be accommodated.
retired:
- word: sequence
became: campaign
resolves_to: nothing
why: >
Renamed, not aliased. Accepting it on input would let stale plans keep working and keep the
old vocabulary alive indefinitely, in the one layer whose job is to fix vocabulary.
- word: metrics
became: engagement
resolves_to: nothing
why: >
The entity collapsed rather than moved: what were separate per-scope reads became one
measure set with the grouping as an argument. There is no one operation the old word maps
onto, so no alias could be honest about what it points at.
concepts:
- id: campaign
prefLabel: campaign
altLabels: [cadence, drip, play, programme, program, nurture track, touch pattern, outreach flow, flow, journey]
hiddenLabels: [campaigns, campain, campagne, cmp, outbound programme, multi-step programme]
definition: >
A named multi-step programme of touches to named people, with its own steps, schedule,
senders, pacing, reply policy and terminal condition. It is always created as a draft;
going live is one separate named act, and it is the only entity in this contract that can
cause sending to begin on its own.
never_means:
- context: marketing
sense: >
A budgeted media effort measured in impressions, reach and spend against a placement —
the sense of the word in advertising, brand and demand-generation work.
because: >
No such object exists in this contract. There is no budget line, no impression, no
placement and no audience bought in aggregate. Every touch here is one message to one
named person on one channel, and the countable things are people, sends, replies and
the meters an operation consumes. A plan that treats a campaign as a spend decision
will look for controls that are not here and will mis-read the ones that are.
- context: political
sense: A time-boxed effort to win something, with volunteers and a deadline.
because: >
Harmless in conversation and misleading in a plan: it carries no notion of steps,
schedule, senders or per-person state, which are the only things a campaign has here.
- id: enrollment
prefLabel: enrollment
altLabels: [enrolment, participation, campaign membership, engagement in a campaign]
hiddenLabels: [enrol, enroll, enrolled, enrollee, participant]
definition: >
One contact's participation in one campaign: its position, state, first-touch timing, the
reason it ended and who or what ended it. It is an entity in its own right — the same
person may participate in several campaigns, and the same campaign holds many
participations, each with its own state.
never_means:
- context: consumer_marketing
sense: A subscription to a list, which the person themself opted into.
because: >
A participation carries no permission of its own and creates none. Permission is
evaluated per contact per channel at send time, by `outreach.precheck`, against
suppression, lawful basis and the channel's own rules. Reading an enrolment as a
subscription is how a suppressed person gets contacted by a plan that looked correct.
- context: hr
sense: Signing an employee up for a benefit or a course.
because: >
The word's everyday sense implies the enrolled party consented and can withdraw. Here
the operator enrols, and the person's own stop is a different act entirely, recorded by
`optout.record` and enforced by suppression.
- id: step
prefLabel: step
altLabels: [touch, touchpoint, message step, wait step, branch, beat]
hiddenLabels: [steps, stp, position, node, action in a campaign]
definition: >
One position in a campaign's ordering: its channel, kind, execution mode, delay with a unit
and a basis, content binding, threading, and for a branch its condition and evaluation
window. Steps are ordered relative to one another, never dated — the calendar is resolved
per person from the schedule, the windows and the blackouts.
never_means:
- context: pipeline
sense: A stage a deal sits in — discovery, proposal, negotiation, closed.
because: >
A step is a position in a programme we control; a pipeline stage is a claim about where
a deal actually is. They move for different reasons and are read by different people.
Stage values are account vocabulary, read with `vocabulary.list`.
- context: workflow
sense: A node in an automation graph that may fan out, loop or call another workflow.
because: >
Steps here are a sequence with declared branches, not a general graph. There is no loop
and no re-entry, because a programme with no terminal condition fails validation.
- id: contact
prefLabel: contact
altLabels: [person, prospect, recipient, human, target]
hiddenLabels: [contacts, cntct, individual, record, people]
definition: >
A person we hold in the working database, with identity keys, fields, provenance, consent
state and history. This is the unit that permission attaches to and that suppression bars —
not an address, not a profile, and not a company.
never_means:
- context: crm
sense: >
A lead — an inbound arrival that has not been qualified or assigned yet, distinct from
a contact attached to an account.
because: >
An inbound arrival is its own entity here, `inbound_lead`, because it has its own clock
and its own first-response obligation. Collapsing the two loses the clock, which is the
only part of it with a deadline attached.
- context: pipeline
sense: The opportunity or deal the person is attached to.
because: >
An opportunity is a separate entity with its own stages and its own owner. A person can
be on several, or none, and their contactability does not follow from either.
- context: telephony
sense: An entry in an address book, identified by a number.
because: >
Identity here is not a single identifier. A person may hold several addresses, numbers
and profiles, each with its own reachability and its own suppression state.
- id: sender
prefLabel: sender
altLabels: [sending capability, sending identity, connected account, channel account, seat]
hiddenLabels: [senders, sndr, mailbox, from-address, sending account, outbound account]
definition: >
A capability to send on one channel — a mailbox at a mailbox provider, an account on a
professional network, a number, or a business messaging identity — together with its
authorisation, warm-up position, configured pacing, pool membership, allowance and the
channel's own sanction on automation. If it cannot send today, that is a fact about the
sender, and it is read before anything is planned against it.
never_means:
- context: message_header
sense: The From line of a message — the display name and address a recipient sees.
because: >
That is one attribute of a sender, and the least consequential one. Everything that
decides whether a message can go out — authorisation, warm-up, restrictions, remaining
allowance, the channel's sanction — sits on the capability, not on the header.
- context: deliverability
sense: The sending domain, its authentication records and its reputation.
because: >
A domain is a separate entity here and is shared by many senders. A healthy domain does
not make a particular mailbox able to send, and one restricted mailbox does not
condemn the domain.
- id: candidate
prefLabel: candidate
altLabels: [found person, discovered person, search result, potential contact]
hiddenLabels: [candidates, cand, lookalike, hit, match]
definition: >
A person located in a data source we do not own and have not taken ownership of. Reading
detail about one, and taking ownership of one, are separately metered; promotion is what
turns a candidate into a contact, and only then does the person exist in our database with
provenance and consent state of their own.
never_means:
- context: recruiting
sense: A job applicant, or somebody being considered for a role.
because: >
Nothing in this contract concerns hiring. The word is used in its literal sense —
somebody who is a candidate for becoming a contact — and the recruiting reading imports
an entire vocabulary of stages and interviews that has no counterpart here.
- context: politics
sense: Somebody standing for election.
because: Listed only because the word is common enough to mis-trigger a resolver.
- id: company
prefLabel: company
altLabels: [firm, business, employer, target company]
hiddenLabels: [companies, co, corp, organisation record, org record]
definition: >
A firm as it exists in somebody else's database — found, resolved or enriched before we own
anything about it. It becomes an `account` the moment it is a record of ours with owners,
state and a working history.
never_means:
- context: crm
sense: The company record we already hold and work.
because: >
That is `account`, and the two carry different obligations: a company is bought or
matched, an account is owned, stated and reported on. `company.resolve` answers "is this
the same firm"; `account.update` changes what we have decided about one.
- id: account
prefLabel: account
altLabels: [organisation, organization, buying organisation, target organisation]
hiddenLabels: [accounts, acct, org, logo]
definition: >
A company as a record of ours: owned, stated, worked and reported on, with members, a state
and a tier.
never_means:
- context: authentication
sense: A login, a seat, or the credential a run authenticates with.
because: >
That is a `sender` or a credential, and confusing them puts a person's mailbox and a
target company in the same word.
- id: engagement
prefLabel: engagement
altLabels: [outreach performance, campaign performance, engagement figures]
hiddenLabels: [engagements, stats, numbers, metrics_view, performance figures]
definition: >
What happened to outreach, read as one measure set over a stated window with the grouping as
an argument — sent, delivered, opened, replied, and every did-not-happen count beside them.
never_means:
- context: reporting
sense: A separate read per scope — one for the account, one per campaign, one per channel.
because: >
The `metrics` entity was retired precisely because it collapsed: there is one
`engagement.summarize` and the grouping is an argument. Looking for a per-campaign
operation is how a name that exists in a single playbook gets invented.
- id: consent
prefLabel: consent
altLabels: [permission, lawful basis, legal basis, opt-in]
hiddenLabels: [consents, gdpr basis, permission record, subscribed]
definition: >
The evidence that reaching a person is permitted: kind, capture artefact, collecting legal
entity, jurisdiction, channels, purposes, subject-matter scope and expiry. Never the answer
on its own.
never_means:
- context: crm
sense: A boolean on the person — consented, yes or no.
because: >
A basis is not a boolean (`A19`). A flag with no artefact behind it is exactly the answer
that cannot be used, which is why `consent.prove` must be able to answer `unprovable` and
why there is no `consent.check`.
- context: marketing
sense: The absence of an opt-out.
because: >
Not having said stop is not permission. An opt-out is an obligation recorded against a
person; consent is evidence recorded about a permission, and neither implies the other.
- id: signal
prefLabel: signal
altLabels: [observation, buying signal, intent signal, trigger event]
hiddenLabels: [signals, sig, intent data, event]
definition: >
One observation about a person or a company, carrying its source, the time it was observed,
a confidence and a validity window.
never_means:
- context: sales_intelligence
sense: A score — an intent number, a fit grade, a heat rating.
because: >
A signal is an observation, never a score (`J1`). A number with no source, no
observation time and no validity window cannot be reasoned about, and acting on a stale
observation is worse than not acting at all (`J2`).
- id: suppression
prefLabel: suppression
altLabels: [exclusion, do-not-contact entry, never-contact, exclusion list entry]
hiddenLabels: [suppressions, dnc, blacklist, blocklist, deny list, suppress, do not mail]
definition: >
A standing bar on contacting a person, an address, a number, a profile, a domain or a whole
company, carrying a reason class and the evidence required to lift it. It is checked before
every send, it survives merges and imports, and no operation anywhere clears one by editing
a field — the only lift is the named one, and it is refusable while a retention floor
applies.
never_means:
- context: deliverability
sense: >
Throttling, deferring or pausing sending — holding traffic back to protect a domain or
a mailbox.
because: >
Those are pacing and incident decisions, reversible by design and taken on our own
behalf. A suppression is a standing prohibition, usually taken on somebody else's
behalf, and treating it as a throttle is how a bar gets lifted when volume recovers.
- context: data_management
sense: Hiding or filtering a record from a view.
because: >
Suppression is not visibility. A suppressed person stays fully readable and fully
reportable; what changes is that they may not be contacted.
- id: disposition
prefLabel: disposition
altLabels: [outcome, business outcome, participation outcome, result]
hiddenLabels: [dispositions, disp, dispo, close reason, end state]
definition: >
The business outcome of one participation, drawn from the account's own declared
vocabulary — what came of it, decided by a human or by a rule, and recorded against the
participation rather than against the person.
never_means:
- context: conversation
sense: What a reply meant — interested, not now, wrong person, please stop.
because: >
That is a classification, on a different axis, recorded by `conversation.classify` with
its own evidence and confidence. Keeping them apart is load-bearing: a request to stop
is an obligation, and if it sat on the outcome axis a relabel would lift a legal duty.
- context: call_center
sense: >
The code an agent or a dialler writes at the end of a call — busy, no answer, left
voicemail.
because: >
That is the outcome of one activity, not of a participation. A call that reached nobody
says nothing about what came of the person's involvement in the campaign.
- context: execution_state
sense: Where the participation currently stands — active, paused, finished.
because: >
State is what the system knows; disposition is what a human concluded. A finished
participation with no disposition is a normal and common condition, and inferring one
from the state would manufacture a business judgement nobody made.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |