Private preview

Build with
Clarity.

Myzelium turns meetings, decisions, specs, code, and tests into one living graph — source-backed context your team and its agents build from.

Access is invitation-only today. Joining the waitlist puts your team in line for the next cohort.

The evidence

A graph of specs, not a folder of files.

We benchmarked Myzelium's spec graph against four file-based spec-driven development frameworks — GitHub Spec Kit, BMAD-METHOD, GSD, and Kiro — on real change requests in two production codebases, with the same model driving every run.

0
source files read per change
Myzelium answers impact questions from the graph
8k–135k
tokens of code read per change
file-based frameworks, growing with change size
1–2
graph queries per change
flat cost, however large the codebase
9
live framework runs
4 scenarios · 2 production codebases
Where specs live
File-based SDD
Markdown files scattered through the repo
Myzelium
A typed graph — states, owners, relationships
“What does this change affect?”
File-based SDD
grep, then read enough code to be sure
Myzelium
One impact query over typed edges
Rules that aren’t in the code
File-based SDD
A dead end — nothing to find
Myzelium
Retrieved from the spec that states them
Prior work with acceptance criteria
File-based SDD
Reconstructed from source, when it exists
Myzelium
Returned as linked sibling specs
Cost as the product grows
File-based SDD
Grows with the codebase
Myzelium
Flat — one or two queries per change

Two kinds of truth

Code carries reality

What runs today. Reading source catches drift, dead wiring, and unfinished substrate — file-based tools are genuinely good at this.

The graph carries intent

Rules, decisions, and acceptance criteria that may not be in the code at all. In our benchmark, one governing rule existed only in the spec graph — no amount of code search could surface it.

Myzelium keeps both connected: agents read your code and query the graph, so intent and reality stay in the same conversation.

Internal benchmark, July 2026 — four change-request scenarios, two production codebases, the same model driving every framework in isolated worktrees. Kiro modeled, not executed. Full methodology available on request.

The substrate

Inside the spec graph.

Every spec-driven framework we surveyed stores specs the same way: flat markdown files in a repo. Myzelium promotes them to typed nodes in one persistent graph — and that single decision is where the benchmark numbers come from.

get_spec_impact("CAP-003")
→ 6 affected specs · composed brief, ~2.4k chars · 0 files read
CAP-003Access controlDM-002User · RoleFEAT-014Role assignmentFEAT-021Module gatingCAP-002Module registryFEAT-009Audit logTS-108Permission testsCAP-007BillingFEAT-030ReportsDM-005Invoice
affected specgenerated test specunrelated — stays out of context

One query, a typed-edge traversal, a composed brief — the round trip an agent makes before every change.

Typed edges

Contains, relates to, depends on — relationships are data you can query, not prose you have to read.

Lifecycle states

Draft, approved, implementing, tested — every spec knows where it stands, and drift gets flagged.

Impact queries

“What depends on this?” is one call over the graph — the blast radius of a change, before you make it.

Composed briefs

Agents receive a few thousand characters of exactly-relevant context — sources, criteria, prior art — never whole documents.

Swipe to compare all five

CapabilitySpec KitBMADGSDKiroMyzelium
Specs as a typed graph, not flat filesNot documentedNot documentedNot documentedNot documentedYes
“What depends on this?” as a single queryNot documentedNot documentedNot documentedNot documentedYes
One persistent graph across the whole productNot documentedNot documentedNot documentedNot documentedYes
Spec health — drift, orphans, lifecycle statesNot documentedPartialNot documentedNot documentedYes
Traceability from spec to code, commit, and testNot documentedNot documentedNot documentedNot documentedYes
Live retrieval over MCPNot documentedNot documentedNot documentedYesYes

The trade, honestly: flat files are zero-setup and review like code in a pull request. The graph is a running service — heavier to adopt, and it pays back as the product and the number of agents touching it grow.

Marks reflect each framework's primary documentation, July 2026 — a dash means not documented, not impossible. Kiro's MCP retrieval is genuine prior art, and we say so.

The system

What Myzelium is.

A living spec graph

Meetings, docs, decisions, code, and tests become source-backed specs — each one knowing its sources, state, and downstream effects.

Humans approve truth

Clarity drafts, flags contradictions, and proposes work. Nothing becomes the source of truth until a person approves it.

Agents build from it

Coding agents receive composed briefs — impact, acceptance criteria, prior art — and ship changes traceable back to their specs.

Be first in line.

We're onboarding teams in small cohorts. Tell us where to reach you.