Glyph 0.3.0 is live — audit gates, glyph_seal, Whyboard + Story templates

[seanhanca]

v0.3.0 is live — the "best-in-class" release. Five PRs closed five sharp gaps that separated Glyph from "another chart library." All backward-compatible with v0.2.0.

📝 Full essay: Charts for agentic frameworks (0.3.0)
📦 Release notes: https://github.com/seanhanca/glyph/releases/tag/v0.3.0

What shipped

Capability	Why it matters
5 new audit rules (AUDIT-05, 12, 13, 14, 15)	Line-on-categorical, too-many-arc-slices, bar-on-quantitative, overlay overload, missing title — misleading patterns the v0.2.0 linter missed. Total: 16 rules.
glyph_seal	Standalone provenance-seal companion to glyph_verify. Emits SHA-256 over (spec, rows, schema) WITHOUT rendering the SVG — useful when a verifier already has the bytes or you want hashes for a content-addressable cache.
4 Whyboard templates	Canonical diagnostic layouts: revenue-miss, conversion-drop, latency-spike, churn-spike. Wired into the playground as a new Whyboard category.
3 Story templates	Canonical 4-panel narrative storyboards: quarterly-review, incident-retro, A/B test readout. New Stories category in the playground.
Audit-aware glyph_spec_patch	Closes the loop on agentic chart refinement. A patch that introduces a new HIGH-severity audit finding is refused with error: "audit_regression" unless the caller passes acknowledged: true.

Counts after 0.3.0

	v0.2.0	v0.3.0
MCP verbs	52	53
Mark types	21	24
Audit rules	11	16
Tests passing	819	1,022

Try it now (no install)

The playground supports ?example=<id> deep links. Click any of these to land directly on a 0.3.0 template:

• 🔍 Whyboard · revenue miss
• 🔍 Whyboard · latency spike
• 📊 Story · quarterly review
• 📊 Story · incident retro

Install

npx -y @glyph/mcp@0.3.0

Or in package.json:

{
  "@glyph/core": "^0.3.0",
  "@glyph/mcp": "^0.3.0"
}

What's next

The five-PR push was a "fill the gaps" release — every PR closed a specific weakness in the agent-stack story. Next on the roadmap (in priority order):

1. MCP-side example flows — agent recipes that chain glyph_render → glyph_audit_spec → glyph_spec_patch → glyph_seal so callers can copy-paste a working end-to-end loop.
2. Whyboard / Story template variants — same shape, different domains (e.g. ops incident vs. business-metric, weekly retro vs. quarterly).
3. More audit rules — file an Issue if you've spotted a misleading pattern Glyph should catch. Each rule is a small PR.

If you build agent stacks, the new verbs are glyph_seal and the updated glyph_spec_patch. Drop a thread under Q&A if you hit anything unexpected, or under Show and tell if you ship something with it — there's a companion thread in Show and tell with use-case demos that I'd love yours added to.

🤖 The five PRs were written, reviewed, and merged by Claude. Same loop the library is designed to support is the loop that maintains it.