Skip to content

[Rewrite P0] Protocol currency audit: NIP registry vs nostream at capture #861

Closed

Description

@chappie-daemon

Parent: #860
Milestone: Phase 0: Protocol currency audit
Host: chappie · Expected cost: $80

Diff the Nostr protocol as of the capture date against what nostream implements, and produce the authoritative NIP list the rewrite must satisfy.

  • Fetch the NIP registry (noscl/nips repo) as of 2026-10-15T00:00:00Z; list every NIP, its status (final/draft/obsolete) and one-line summary
  • Diff against supportedNips (25 NIPs: 1,2,3,4,9,11,12,13,14,15,16,17,20,22,25,28,33,40,42,43,44,45,50,65,77) — for each gap decide: must-have for parity, opportunistic add, or out of scope; record the decision
  • Check NIPs that changed since nostream implemented them (e.g. NIP-11 metadata fields, NIP-42 auth flow, NIP-77 negentropy status) — flag behavioural deltas with NIP-spec citations
  • Check the 24 post-capture commits (through 2026-10-15T00:00:00Z) for protocol-affecting changes (measured: NIP-77 wire protocol landing in 7409006/7409006-class commits, settings URL validation 48fe9c4)
  • Output: docs/rewrite/nip-parity.md in the repo — table of NIP → status in nostream → delta → rewrite disposition; the disposition rows become issues in Phase 2 or explicit "not implementing" notes

Acceptance: every one of the 25 advertised NIPs has a disposition; every registry NIP marked final/active has a decision; the deltas carry citations to the NIP text.

Activity

  1. chappie-daemon commented on Oct 17, 2026

    @chappie-daemon
    CollaboratorAuthor

    Outcome — escalated, awaiting review on the PR.

    1. Escalated at docs: add NIP parity audit for the rewrite #872 (docs-only, empty changeset, not merged). Rounds 1–4 were reviewed — 22 lens runs, 32 findings raised and closed; round 5 built the response to round 4's eight findings and no lens reviewed that head, so the PR is one round ahead of its last completed review. decision.md in .factory/jobs/001-nip-audit/ records what was and was not reviewed.
    2. Expected $80.00 vs actual $3.71 — 30 agent sessions, 81 min end-to-end, 1,362 requests. Per stage: recon 4.6%, spec 2.4%, build 27.7%, review scoping 5.1%, lenses 60.2%.
    3. Review share: the six lenses cost 60.2% of the job and are where the quality came from. 6 of 22 lens runs passed on first sight; 16 returned CHANGES. Cost per round stayed flat ($0.64 / $0.64 / $0.59 / $0.53) while the document grew 600→840 lines, because each lens got its own prior review plus a named delta instead of the whole file.
    4. Spans-discovery defect: factory-spans.ts counted 24 of 30 sessions ($3.30 of $3.71) — it silently drops agent sessions it cannot resolve to the orchestrator (recon, spec, four scopers here). Corrected figure and method in cost.md.
  2. chappie-daemon commented on Oct 17, 2026

    @chappie-daemon
    CollaboratorAuthor

    Merged as 526054b (#872): docs/rewrite/nip-parity.md, 99 dispositions, 11 delta sections, 11 Phase 2 items. Greptile's three P2 findings were fixed in dc5b32a before the merge (NIP-09 a-tag deletion, NIP-28 kind 41 replaceability, NIP-11 test coverage), threads resolved, oracle PASS. Actual cost $3.71 vs $80 expected (0.046x) — calibration recorded in #860.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Labels

factoryBuilt by the software factoryrewritenostream 4.0 Rust/LMDB rewrite

Projects

No projects

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions