What does this skill do, and when should you use it?
Consent Registry is an email-subscription consent and live-suppression registry skill. It records pseudonymous opt-ins, lawful-basis evidence, unsubscribes, complaints, hard bounces, restores, and erasure events. The authoritative state is the append-only NDJSON event stream and replay query, not a cache or manually edited projection. It does not send email, edit ESP state, or declare a list safe on its own.
Reads memory/events/consent.ndjson, its projection, and minimum proof references; queries live suppression through scripts/registry-events.py; and appends consent, suppress, restore, erase, or propose events. It validates authorization, source, timestamps, basis and proof references, revisions, and idempotency data. It regenerates accepted per-subject views and reports pseudonymous IDs, event IDs/offsets/revisions, current suppression results, missing evidence, and a next-skill handoff. Raw contact PII must not be stored.
- An email-compliance team needs to record a subscriber’s opt-in timestamp, lawful basis, and double-opt-in proof reference.
- An ESP reports an unsubscribe, complaint, or hard bounce and the subject must be suppressed immediately.
- A verified fresh opt-in arrives after withdrawal and an authorized host must restore the subscription.
- A data subject requests erasure and projected payload must be removed while a suppression tombstone remains.
- A sending workflow needs to verify whether a pseudonymous subject is currently suppressed.
How do you install this skill?
- This review covers only protocol/consent-registry/SKILL.md and the shared materials supplied by the user; no scripts were executed, and the referenced runtime, event protocol, schema, or catalog were not verified.
- Before deployment, confirm that host capability keys cannot be exposed to an agent-controlled shell, and separately verify whether erasure covers backups, snapshots, Git history, and exports.
- SECURITY.md lists 17.x as the current supported line while the skill declares 18.0.0; confirm release support and runtime compatibility first.
- Shell / CLI
- Local filesystem
Python 3Git
This skill is part of the aaron-he-zhu/aaron-marketing-skills collection. In Claude Code, run /plugin marketplace add aaron-he-zhu/aaron-marketing-skills, then /plugin install aaron-marketing@aaron. On a compatible generic Agent Skills host, run npx skills add aaron-he-zhu/aaron-marketing-skills. Alternatively, run git clone https://github.com/aaron-he-zhu/aaron-marketing-skills. The source does not document a standalone installation command for this skill.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/aaron-he-zhu/aaron-marketing-skills.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/protocol/consent-registry" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- Record opt-in for subject sha256-7d9f with basis, proof references, and timestamp,
- Immediately suppress sha256-7d9f from unsubscribe webhook evt-882,
- Is sha256-7d9f suppressed right now?
After installing the skill directory in a compatible Agent Skills host, use prompts such as “Record opt-in for subject sha256-7d9f with basis, proof references, and timestamp,” “Immediately suppress sha256-7d9f from unsubscribe webhook evt-882,” or “Is sha256-7d9f suppressed right now?” Mutations require the runtime, the required host capability, and explicit authorization. Eligibility and send queries must run python3 scripts/registry-events.py is-suppressed <subject-id>.
What are this skill's strengths and limitations?
- Uses an append-only event stream and replay query as the source of truth.
- Provides an immediate deny-only path for unsubscribes, complaints, and hard bounces.
- Requires authorization, source, date, and proof references while prohibiting raw contact PII.
- Separates proposals, accepted appends, restores, and erasure safety handling.
- Requires the verified host capability, runtime script, event schema, and system catalog; without them it cannot claim canonical state.
- Records and queries consent and suppression only; it does not send email, edit an ESP, or build segments.
- The source provides no standalone installation procedure or test-coverage evidence for this individual skill.
- The skill frontmatter reports version 19.0.0 while the README collection badge shows 18.0.0, so adopters should verify release synchronization.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Consent Registry this page | 62 · Recommended | ★ 3k | 12d ago | Apache-2.0 |
| Launch Registry | 59 · Recommended | ★ 3k | 12d ago | Apache-2.0 |
| Channel Registry | 58 · Recommended | ★ 3k | 12d ago | Apache-2.0 |
| Creator Registry | 61 · Recommended | ★ 3k | 12d ago | Apache-2.0 |
| Storage Analyzer | 60 · Recommended | ★ 21k | 25d ago | MIT |
The skill explicitly hands SEND scoring to email-quality-auditor and segmentation to list-segment-builder; it is not a replacement for either an email auditor or a segment builder.
How did FollowSkills review this skill?
The skill explicitly limits records to pseudonymous subjects, forbids raw contact PII, requires host-bound authorization capabilities, immediate suppression, append-only writes, erasure tombstones, explicit permission, and live-suppression checks before downstream sending. README.md and SECURITY.md also describe fail-closed authority and recovery boundaries. Deduction of 4: core safeguards depend on the unprovided protocol, runtime, catalog, and host-deployment files; runtime memory is not encrypted, and erasure does not cover backups or historical copies. Publisher identity is unverified but is not deducted by itself.
The skill specifies replay, expected_revision, verification, projection regeneration, proposal fallback, and independent suppression queries, with generally consistent control flow. Deduction of 2: registry-events.py, its schema, runtime, and catalog are not included in the supplied evidence, so executability, edge-case handling, and diagnostic failure feedback cannot be confirmed; static calibration caps this at 10.
Triggers, pseudonymous-subject inputs, consent/suppression event scope, outputs, and exclusions are reasonably clear, and standalone-install degradation is disclosed. Deduction of 4: concrete input schemas, user-interaction examples, and environment/mainland-China reachability evidence are limited; the core capability depends on a local host runtime rather than a network service.
The skill has structured frontmatter, versioning, Apache-2.0 licensing, Quick Start, contract, data sources, numbered instructions, limitations, and next-skill handoffs. README.md documents the shared contract, installation, versioning, and security context. Deduction of 5: no explicit maintainer, changelog, troubleshooting/FAQ, or complete dependency-install instructions are supplied for this skill; referenced materials are absent, and SECURITY.md has a possible 17.x versus skill 18.0.0 support-version mismatch.
For consent logging, immediate unsubscribe suppression, lawful-basis lookup, restoration, and erasure, the skill defines an authoritative event stream, completion conditions, and handoff outputs that are conceptually directly usable. Deduction of 8: there is no verified representative output, execution evidence, or comparison with manual/alternative workflows; static calibration therefore limits effectiveness to 7.
The repository supplies an Apache license, architecture documentation, CI workflows, and tests for architecture, audit artifacts, and PII scanning, providing partial auditability. Deduction of 5: the supplied tests do not cover this skill's key consent-registration paths, referenced protocols and runtime implementation are missing, and there is no third-party execution evidence; static calibration caps verifiability at 5.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →