V VIM
Pre-launch · Targeting OpenSSF Sandbox · Apache 2.0

Mythos finds the zero-day.
VIM maps its blast radius.

Claude Mythos Preview is surfacing critical zero-days faster than any team can track where they live. VIM is the lineage layer that turns a single upstream finding into the full map of every fork, vendored copy, and compiled binary running the same vulnerable code.

The problem

Finding bugs is cheap. Tracking where they live is not.

Frontier models like Anthropic's Claude Mythos Preview have collapsed the cost of finding flaws. The bottleneck now is mapping where the vulnerable code actually lives. No existing tool answers the question: given a vulnerability in project X, which forks, vendored copies, and compiled binaries are also affected?

59.3%

of vulnerable copied files are C code — invisible to package-level SCA tools.

World of Code research, Mockus et al.

17%

of open-source components enter codebases outside package managers entirely.

Black Duck 2026 OSSRA

37 days

median patch-porting delay across hard forks — 25%+ take over 3 months.

Pan et al., ISSTA 2024

Regulatory deadline. EU Cyber Resilience Act takes effect 11 September 2026. Penalties up to €15M or 2.5% of global turnover for non-compliance — and it requires lineage evidence that current SCA tools cannot provide.

The discovery engine

Built for the discovery wave from Claude Mythos Preview.

Anthropic's Claude Mythos Preview is, in its own words, "capable of identifying and then exploiting zero-day vulnerabilities in every major operating system and every major web browser." In a single month it surfaced thousands of high- and critical-severity bugs — among them a 27-year-old flaw in OpenBSD and a 16-year-old vulnerability in FFmpeg.

This is a different kind of tool doing a different kind of work. And finding the bug faster isn't enough — the moment Mythos reports a flaw upstream, the urgent question becomes: where else does this exact code already live? That is the question VIM exists to answer.

1 · Discovery

Mythos · Project Glasswing

Finds the zero-day in upstream code.

2 · Lineage

VIM

Traces every fork, vendored copy, and modified variant carrying the same code.

3 · Remediation

GUAC · OSV · deps.dev

Maps affected artifacts and drives the fix downstream.

How it works

Trilateral identity resolution.

VIM identifies every code unit three ways at once. Combined, they survive renames, vendoring, forks, and non-trivial edits.

Extrinsic

pURL

ECMA-427 package identity. Ecosystem-aware. Aligned with OSV, GUAC, and deps.dev.

pkg:npm/lodash@4.17.21
Intrinsic

SWHID

Software Heritage ID. Content-addressable. Identifies code by what it is, not what it's called.

swh:1:cnt:3b18e512...
Fuzzy

TLSH

Locality-sensitive hash. Survives patches and refactors. Similarity becomes a distance query.

5ca1d8a25f213672...

Prototype signal

Measured TLSH distance in our end-to-end test harness:

Relationship TLSH distance
Byte-identical 0
Single-site security patch 48
Unrelated code 297

Runnable: go test ./pkg/identity/

Ecosystem

Designed to complement, not replace.

VIM is the lineage layer between upstream discovery and downstream remediation. It is built to interoperate with existing OpenSSF projects — not duplicate them.

Existing tool What it does How VIM extends it
GUAC Artifact composition Adds fork lineage + code-level provenance
OSV Vuln-to-version mapping Traces vulns across forks and copies
deps.dev Dependency resolution Tracks vendored code outside dep trees
Scorecard Security posture Monitors patch propagation across forks
SLSA / Sigstore Build integrity Connects binary provenance to source lineage

Data sources

Software Heritage World of Code OSV.dev deps.dev GUAC CISA KEV Sigstore / Rekor SBOM (SPDX, CycloneDX)

Try the prototype

From a Mythos finding to its blast radius — today.

The trilateral identity layer and an in-memory inheritance graph are runnable right now — including ingesting a Mythos report and tracing a finding across a corpus of forks and copies. The persistent graph backend (JanusGraph + Cassandra) and GUAC integration land later in the roadmap.

bash
# Clone and build
git clone https://github.com/abduljaleel/vim.git
cd vim
go build -o bin/vim-cli ./cmd/vim-cli

# Compute a LineageID for any file
./bin/vim-cli identity README.md 'pkg:github/abduljaleel/vim@pre-alpha'
SWHID: swh:1:cnt:5049b7304c2b1037...
pURL:  pkg:github/abduljaleel/vim@pre-alpha
TLSH:  5ca1d8a25f2136724b37cecb...

# Compare two files by similarity
./bin/vim-cli similarity a.c b.c
TLSH distance: 48
Result: MODERATE similarity (possibly related / patched variant)

# Trace a Mythos finding across a corpus of forks & copies
./bin/vim-cli query examples/mythos/findings.json examples/mythos/corpus
CONFIRMED · byte-identical copies of the vulnerable code (2):
  - upstream/parse.go
  - verbatim-fork/parse.go
REVIEW · shares lineage, fix may or may not be applied (2):
  - patched-fork/parse.go      TLSH distance 25
  - cosmetic-fork/parse.go     TLSH distance 26

Roadmap

Launch to production.

  1. 1

    Foundation · Apr–May 2026

    OpenSSF Sandbox submission · Alpha-Omega & CRC-P grant applications · Community engagement at OpenSSF Community Day NA.

  2. 2

    Technical build · Jun–Jul 2026

    JanusGraph + Cassandra core · Software Heritage ingestion · Phase 1 GUAC integration · Mythos / Project Glasswing vulnerability report tie-in.

  3. 3

    Community & market · Aug–Oct 2026

    Alpha release at DEF CON 34 · EU CRA enforcement window outreach · OpenSSF Incubation application · Beta release with Phase 1 GUAC integration complete.

Get involved.

VIM is pre-launch. We are recruiting maintainers, research partners, and design-partner organisations with real supply chain compliance problems.