# Design provenance and release boundary

## Origin

Michael Briffa's recognition conjecture and his iterative edits to private operator-research outreach informed this skill: individual niche contribution; warm curiosity; transparent identity and research purpose; explicit why-you connection; concrete discussion topics; one clear ask; complete-email review; preservation of approved shared copy.

The public bundle contains fictional teaching examples. It excludes the original contact list, inbox records, recipient addresses, private correspondence, session transcripts, internal paths, and send receipts. No response-rate or conversion result is claimed.

## Discovery and selection

Inspected through Hermes's native skills.sh discovery on 2026-10-14:

1. [gtmagents/gtm-agents — cold-email-personalization](https://skills.sh/gtmagents/gtm-agents/cold-email-personalization). Repository: https://github.com/gtmagents/gtm-agents . Repository license inspected: Apache-2.0. Read the complete SKILL.md and all ten linked Markdown assets. This is a sales-campaign workflow with research, variable management, drafting, scoring, and sequences. Its useful design primitives were source-grounded research, explicit QA, and reading from the recipient's perspective. Its numeric scoring thresholds, word-count prescriptions, sales-offer defaults, follow-up cadence, and role-based motive assumptions were not adopted.
2. [coreyhaines31/marketingskills — cold-email](https://skills.sh/coreyhaines31/marketingskills/cold-email). Repository: https://github.com/coreyhaines31/marketingskills . Repository license inspected: MIT. Read the complete SKILL.md and the personalization reference; other linked modules were outside the adopted drafting scope and were not relied on. Useful design primitives: facts before angle, connection between the detail and purpose, natural voice, and permission to find no hook. Did not adopt the unsourced-in-this-build performance figures, secondary-domain setup, paid-tool integrations, sending, cross-channel automation, or generic segment fallback as a successful personalized result.

Decision: reference-only, independently written procedure. No community code or package was installed or executed; no upstream prose or examples are redistributed. The skill has no runtime dependencies beyond an agent able to read Markdown and use supplied material. Research tools are optional host capabilities. No dependency on Michael's private skills or environment is included.

## Release status

v0.2.1 is the public MIT-licensed release authorized by Michael Briffa on 2026-10-15. The v0.2.0 Draft and Check procedures are unchanged: confirmed sender context, individual source research, and sample approval before batch expansion. The change is public-release metadata, documentation and licensing. See the included LICENSE.txt. No community prose or executable code is redistributed.