brand-enforcement

name	brand-enforcement
description	Brand voice checklist for every external-facing content review. Apply during review_content tasks before scoring the submission.
metadata	{"version":"1.0.0","type":"always-active"}

Brand Enforcement

The canonical brand voice is defined in prompts/brand-voice.md. This skill is the operational checklist Reviewer runs on every submission.

---

Voice — Four Attributes

Every piece of content must be all four:

Attribute	Sounds Like	Doesn't Sound Like
Direct	"12 agents. 6 departments. Zero humans in the loop."	"We're excited to announce our innovative AI-powered solution…"
Technical but accessible	"Event-driven architecture means agents react to what happens, not what's scheduled."	"Our proprietary synergistic platform leverages cutting-edge paradigms…"
Confident	"Open source because agent infrastructure shouldn't be a walled garden."	"The ONLY solution for enterprise AI orchestration!"
Community-first	"Fork it. Break it. Ship something better. That's the point."	"Join our exclusive early access program…"

If the content fails any attribute → reduce score to <90. If it fails two or more → reviewer:flagged with specific rewrite suggestions.

---

Banned Terms (automatic reduction)

The following marketing-speak is banned. Flag and rewrite.

Banned	Why	Use Instead
"revolutionary"	Empty superlative, hype-marketing	Describe what it does
"game-changing"	Same	Describe the actual change
"cutting-edge"	Cliché	Name the specific technology
"next-gen"	Lazy	Name the generation if relevant
"synergy" / "synergistic"	Corporate filler	Remove
"leverage" (as verb)	Vacuous	"use"
"solution" (generic)	Meaningless	Name what it is (framework, library, agent)
"AI-powered" (as differentiator)	Redundant for this product	Name the specific capability
"We're excited to…"	Cliché opener	State the thing directly

---

Required Framing

Public-facing content MUST use this framing order:

1. What it does (one concrete sentence)
2. How it works (one sentence on mechanism)
3. Why it matters (one sentence on consequence)

NOT the reverse. Benefit-first copy reads as marketing. Mechanism-first reads as engineering confidence.

Example

Bad (benefit-first):

"Transform your organization with next-gen AI orchestration that unlocks unprecedented productivity."

Good (mechanism-first):

"YClaw runs agents in real departments with event-driven coordination and approval gates. Extracted from a production system that ran 12 autonomous agents for over a year. Open source under AGPL-3.0."

---

YClaw Capitalization (strict)

Form	Where it's correct
YClaw	Default for all external / user-facing content — name of the product
YCLAW	Repo name + internal documentation + path names only (e.g., YClawAI/YClaw)
yclaw / Yclaw / YClaw™	❌ Never

Fail-closed rule: if you see YCLAW in user-facing content, flag it. If you see yclaw or Yclaw anywhere, flag it.

The package / GitHub repo name is YClaw. The brand in prose is YClaw. The internal long-form / repo identifier is YCLAW.

---

Features: Present Tense Only

Never use future tense for shipping announcements. If it ships, it IS shipped, not "will be shipped":

• ❌ "YClaw will support multi-model LLM providers."
• ✅ "YClaw supports Anthropic, OpenAI, Google, and OpenRouter."

If a feature is NOT shipped:

• ❌ "Coming soon: feature X"
• ✅ "Feature X is tracked on the roadmap. Not shipped yet."
• Or just don't mention it.

---

Links Policy

Every external-facing post MUST include at least one of:

• https://github.com/YClawAI/YClaw
• https://yclaw.ai
• A specific deep link to docs or an issue

Links must be:

• HTTPS only
• Not tracking-wrapped (no ?utm=* unless Reviewer/Ember explicitly approves)
• Not shortened (no bit.ly, no tinyurl) unless it's a Slack/Discord limit
• Pointing at canonical YClawAI/YClaw (not forks, not mirrors)

---

Legal Rails

From prompts/brand-voice.md:

• NEVER use: Deceptive capability claims, fabricated benchmarks, impersonation of other projects, unsubstantiated superiority claims.
• NEVER reference: Internal infrastructure details, API keys, credentials, other organizations' proprietary information.
• Safe topics: Open source, AI agents, infrastructure, developer tools, community building.

---

Voice Score Rubric

Apply this to every review:

Score	State
90–100	All four voice attributes, no banned terms, framing correct, links present
70–89	One minor voice slip OR one stylistic issue (e.g., missing link, mild hype)
50–69	Two+ voice issues OR one structural issue (framing reversed, YClaw capitalized wrong)
0–49	Banned term used, future tense on shipped feature, legal-rail violation

Anything ≤49 → automatic BLOCK with specific rewrite. Legal flag at any score → automatic BLOCK regardless of voice.

---

Out of scope

• Securities / financial claim screening → see claims-risk skill (separate, higher priority).
• Per-channel format rules → see channel-standards skill.
• License / attribution language → see oss-legal-guardrails skill.