← The register

FIGURE · repos-public

Public repositories

48 (116 total owned)

How it was taken
Public repository count from the GitHub GraphQL API: owned, non-fork repositories filtered to public visibility. The total owned non-fork count comes from the same query.
Measured
2026-10-30
Reproducible
Yes: clone and re-derive.

What the number is made of

  1. The repository list

    Every public repository on the account, openable in one page.

    Open the reference →
  2. operant

    The Rust agent runtime: 7,168 tests passing at its pinned gate commit.

    Open the reference →
  3. holon

    The agentic-OS kernel and the flagship of the portfolio, public since 2026-10-25.

    Open the reference →
  4. switchpost

    The social posting engine, still named social-forge on GitHub: 389 MCP tools, countable in the tree.

    Open the reference →
  5. deckmill

    Browserless slide factory in Rust: 46-entry slide registry, 192 tests at its gate commit.

    Open the reference →
  6. scorestrata, memory-wire, project-genome

    The deterministic music compiler, the memory-bank CLI, and the repository governance tooling, all public with executed gates in this register.

    Open the reference →

How to check it

The profile is public, so this is checkable directly. The figure has moved repeatedly under measurement: see reconciliations R1, R8, R22 and R24.

Open the artifact

This figure's history

  • R1

    73 → 77

    The register figure was read from the GitHub API on 2026-09-22. The profile page read 73 on 2026-09-23. Both readings counted a moving target, and the dispute stayed open for a month. R8 settled it by re-measuring, which is the only move that resolves a disagreement like this one.

  • R8

    77 public, 138 total (2026-09-22), with R1 open between 73 and 77 → 67 public, 132 total (2026-10-24)

    Re-measured with the GitHub GraphQL API on 2026-10-24. Against the September count the total moved by six and the public count by ten; the register does not attribute the split between visibility changes and deletions, because attributing it without the per-repository log would be guessing. This closes R1 and replaces the September figure everywhere the site prints it.

  • R22

    67 public, 132 total (2026-10-24) → 69 public, 131 total owned non-fork (2026-10-26)

    The freshness ladder in the identity store recorded 69 public on the same evening the register printed 67: the register had picked up the earlier same-day reading and went stale within hours of measuring. Re-measured with the same GraphQL query on 2026-10-26. The lesson: a figure on a moving target is a claim about a moment, and the row now dates the moment it means.

  • R24

    69 public, 131 total owned non-fork (2026-10-26) → 48 public, 116 total owned non-fork (2026-10-30)

    This row printed a figure its own method did not support. The 69 public came from an UNAUTHENTICATED public count, while the 131 total came from an AUTHENTICATED private-plus-public count -- two different instruments, so the pair could never reconcile, and the public half overstated this account by 21. The authoritative profile (identity-core, sub-systems/professional/data/candidate-profile.yaml, measured 2026-10-30 by one authenticated source for all three numbers) now reads 116 total / 48 public, and an unauthenticated public re-check on 2026-11-01 reads 55 public including forks, 48 excluding -- which matches. Superseded rather than deleted, per the register convention. The lesson is sharper than the R22 row held: a figure on a moving target is a claim about a moment (R22), and a figure assembled from two different instruments is not a figure at all.

Where this site cites it