| name | scope-need |
| description | Use when a user brings a new capability, feature request, or enhancement need for host in vague terms. Interviews conversationally and produces a scoped-need document. Applies Gate 1 (host-mission alignment), Gate 2 (narrowest scope), and Gate 3 (specialist routing) before producing output. |
Scope a need
A need arrives fuzzy. A feature arrives sharp. This skill is the conversation that turns fuzzy into sharp, with three specific filters: host-mission alignment, narrowest-scope discipline, and specialist-skill routing.
Your posture
You are interviewing, not pitching. host is a managed-hosting control plane with a governance-first posture, on-chain anchoring, multi-tenant isolation, and a narrow mission: run lesser instances responsibly for customers, anchor agent identity on-chain, underwrite operator trust via attestation + evidence, and enforce operational quality through the gov-infra rubric.
The scoping question is always three-part:
- Is this host-mission work — governance, multi-tenant, on-chain, managed-provisioning, managed-updates, trust-API, consumer-release-verification, comm-routing, or operator-portal — or is it scope growth outside that mission?
- If it's in-mission, what is the narrowest possible scope that preserves governance rubric, multi-tenant isolation, on-chain integrity, consumer release verification, trust-API rigor, CSP, AGPL coverage, and idiomatic framework consumption?
- Does the change touch the governance rubric, provisioning / managed-update, soul registry, trust API / CSP / instance-auth, or framework consumption? If yes, route to the appropriate specialist skill before enumeration.
The default for security, multi-tenant, on-chain-integrity, governance, trust-API, release-verification, operational-reliability, AGPL, and framework-feedback work is "yes, evaluate at Gate 2." The default for net-new capability outside the mission is "no."
Start with memory and the architecture
- Read
README.md, AGENTS.md, gov-infra/README.md, gov-infra/AGENTS.md, and docs/ for canonical architecture and contracts.
memory_recent — has this need or adjacent work been scoped before?
query_knowledge — do AppTheory / TableTheory / FaceTheory or sibling equaltoai repos already cover this concept?
If tools are unavailable, surface and ask the user to re-auth.
The interview
Ask, one or two at a time:
- Who is asking and why now? Customer / operator report, security finding, CVE, compliance requirement, advisor-dispatched brief, or Aron-direct?
- What problem does it solve? Current pain, not speculative improvement.
- Which surface does it touch?
- Control-plane API / portal
- Trust API / attestations / instance-auth
- Soul registry (on-chain + off-chain)
- Provisioning worker / managed-update workers
- AI worker / comm worker / render worker / soul-reputation worker
- Email ingress
- Governance rubric / verifier / evidence
- CDK / IaC / CloudFront / DynamoDB
contracts/ (Solidity) / Hardhat / Safe-ready payloads
web/ SPA / CSP
- AGPL / licensing
- Framework consumption pattern
- Which consumers are affected? Managed-instance operators, customers (portal), sibling repos (lesser / body / soul / greater), on-chain actors, external vendors (Stripe, AI providers), trust-API public readers.
- Is this a tenant-isolation-affecting change? If yes, elevated scrutiny.
- Is this an on-chain or Safe-ready-governance change? If yes, route to
evolve-soul-registry.
- Is this a governance-rubric change? (Adding a verifier, tightening thresholds, changing evidence policy) route to
maintain-governance-rubric.
- Is this a provisioning / managed-update change? (Including consumer release verification) route to
provision-managed-instance.
- Is this a trust-API / CSP / instance-auth change? Route to
audit-trust-and-safety.
- Is this framework-awkward? Route to
coordinate-framework-feedback.
- What does success look like? Observable, testable.
- What is explicitly out of scope?
The three gating questions
Gate 1: Is this host-mission work?
Eight possible verdicts:
- Yes — security / tenant-isolation / on-chain-integrity work. CVE response, isolation bug, on-chain miscalculation fix. Always accepted. Proceed to Gate 2.
- Yes — governance-rubric work. New verifier, tightened threshold, evidence-policy refinement. Proceed; route through
maintain-governance-rubric.
- Yes — provisioning / managed-update work. New provisioning step, managed-update recovery improvement, consumer-release-verification refinement. Proceed; route through
provision-managed-instance.
- Yes — soul-registry work. New contract function, Safe-ready payload pattern, off-chain state migration. Proceed; route through
evolve-soul-registry.
- Yes — trust-API / CSP / instance-auth work. New attestation type, tightened auth, CSP refinement, safety evidence service. Proceed; route through
audit-trust-and-safety.
- Yes — operational-reliability / AGPL / observability work. Latency, availability, observability for observed gaps, rate-limiting, license vetting, CVE-response hardening. Accepted.
- Yes — framework-feedback work. A host concern surfaces AppTheory / TableTheory / FaceTheory awkwardness. Route through
coordinate-framework-feedback.
- No — out-of-scope growth. Tenant-side data operations, tenant user management, tenant content moderation, general identity-provider scope for non-lesser consumers, general payments processing beyond tipping and host billing. Produces a redirect document.
Gate 2: What is the narrowest possible scope?
Prefer:
- Bug fixes scoped to the specific reported symptom
- Governance-rubric additions (new verifier) over modifications (loosening an existing one)
- Additive provisioning steps that preserve existing flows
- On-chain additions (new function on existing contract) over replacements; Safe-ready-governance-coordinated for non-trivial changes
- Trust-API additions (new attestation type) over modifications of existing shapes
- CSP additions (new
'self'-origin surface) over loosening (new third-party origin)
- Dependency bumps within current major versions
Avoid:
- Refactors "while we're in there"
- New verifiers that aren't deterministic or don't produce evidence
- Provisioning-pipeline rewrites; incremental improvements preferred
- On-chain refactors without clear upgrade path
- Trust-API breaking changes
- CSP loosening ever
- Cross-tenant queries in host's DynamoDB
- Reading tenant content into host's plane
Gate 3: Specialist routing
If the change touches any of these, the specialist skill runs before enumeration:
- Governance rubric, verifier, evidence, pack.json →
maintain-governance-rubric
- Provisioning, managed-update, consumer release verification →
provision-managed-instance
- Soul registry (on-chain + off-chain + governance) →
evolve-soul-registry
- Trust API, CSP, instance-auth, attestations →
audit-trust-and-safety
- Framework awkwardness →
coordinate-framework-feedback
- Advisor-dispatched brief →
review-advisor-brief
Output: the scoped-need document
For Gate 1 verdict "host-mission work":
# Scoped Need: <short name>
## Background
<one paragraph>
## Driver
<operator / customer / CVE / compliance / advisor-dispatched / Aron-direct>
## Problem
<what is broken, missing, or painful today>
## Surface affected
<control-plane / trust-API / soul / provisioning / managed-update / worker / email-ingress / gov-infra / CDK / contracts / web / AGPL / framework>
## Lambda(s) affected
<enumerated, or "none — infrastructure / gov-infra / contracts only">
## Classification
<security / tenant-isolation / on-chain-integrity / governance / provisioning / managed-update / soul-registry / trust-API / CSP / operational-reliability / AGPL / framework-feedback / bug-fix / test-coverage / dependency-maintenance / docs>
## Narrowest-scope proposal
<smallest change that addresses the need>
## What this need explicitly does not cover
<bounded scope>
## Success criteria
<observable, testable>
## Specialist routing
- Governance rubric: <not touched / walk via maintain-governance-rubric>
- Provisioning / managed-update / release verification: <not touched / walk via provision-managed-instance>
- Soul registry: <not touched / walk via evolve-soul-registry>
- Trust API / CSP / instance-auth: <not touched / walk via audit-trust-and-safety>
- Framework consumption: <idiomatic / awkwardness via coordinate-framework-feedback>
- Advisor brief: <n/a / review via review-advisor-brief>
## Consumer impact
<managed operators / customers / sibling repos / on-chain actors / external vendors>
## Multi-tenant isolation impact
<none / elevated scrutiny required — document>
## On-chain impact
<none / off-chain only / on-chain — Safe-ready path>
## AGPL posture
<no change / confirmed AGPL-compatible / decision required>
## Open questions
<unresolved>
For Gate 1 verdict "out-of-scope growth":
# Redirect: <short name>
## Background
<what was asked>
## Why this doesn't belong in host
<scope bounded to managed-hosting control plane; this is X, which belongs in Y>
## Appropriate owner
<tenant-side lesser / body / soul / greater / Theory Cloud framework / separate repo / scoping with Aron>
## Path for the requesting user
<rough outline>
## Recommended next step
<specific handoff>
Persist before handoff
Append only when scoping surfaces a recurring pattern — a redirect category, a governance-rubric evolution signal, a multi-tenant-isolation pattern, an on-chain coordination pattern. Routine completions aren't memory material. Five meaningful entries beat fifty log-shaped ones.
Handoff
- In-mission, specialist walk required — invoke the appropriate specialist skill before enumeration.
- In-mission, none of the above — invoke
enumerate-changes directly.
- Advisor-dispatched scope —
review-advisor-brief already ran; output includes Aron's authorization.
- Out-of-scope — redirect document is the handoff.
- Resolved to "no change needed" — record and stop.
- User defers — record and stop.