v3-spine-brief

v3 spine brief: shared frame for fanout drafters

Metadata

Statusdone
Assignedagent-59
Agent identity3184716484e6f0ea08bb13539daf07686ee79d440505f1fdf2de0357707034c3
Created2026-05-01T20:42:55.231091781+00:00
Started2026-05-01T20:45:53.411933585+00:00
Completed2026-05-01T20:49:05.791218622+00:00
Tagsgrant,urgent,v3,spine, eval-scheduled
Eval score0.83
└ blocking impact0.95
└ completeness0.98
└ constraint fidelity0.55
└ coordination overhead0.95
└ correctness0.95
└ downstream usability0.95
└ efficiency0.90
└ intent fidelity0.78
└ style adherence0.98

Description

Description

Erik is reframing the Google.org application via fanout/fanin. Five parallel drafting agents will each refit different sections to the new spine. They need a shared one-page reference so they don't drift away from each other. This task produces that reference.

This is the FIRST task in the v3 fanout. It blocks all five drafters and the assembler.

Output: ~/poietic.life/notes/v3-spine-brief.md (max 500 words, terse).

What to read

  1. workgraph_google_application_FINAL_v2.md (post-audit-fix state) — to see what's being replaced
  2. STATE.md §5 (decision log; note that v2 itself is now being superseded by v3)
  3. CLAUDE.md 'Key Narrative Decisions' and 'Style Preferences'
  4. ~/poietic.life/index.html for the verbatim PBC public benefit statement
  5. The Delaware Certificate of Amendment at ~/poietic/corporate/signed/amendments/01_Certificate_of_Amendment.pdf.txt if the verbatim wording isn't in the landing page

What to write

A short doc with these labeled sections:

1. SPINE (≤80 words)

The one-paragraph thesis statement of the v3 application. Anchor: Poietic PBC builds open-source infrastructure (WorkGraph) for reliable, careful, auditable hybrid human-AI work in computational and clinical genomics. The PBC mission ('make human and machine collaboration legible and responsive to its participants') is the spine; the science is the proving ground.

2. CONCRETE ANCHOR: REFERENCE RESOURCES (≤80 words)

Erik's authoritative track record includes building foundational REFERENCE RESOURCES (vg, PGGB, HPRC's human pangenome reference, comparative pangenome references). These are very high stakes: errors and biases propagate to thousands of downstream labs. 'Get the reference correct' is the operational form of 'reliable, careful, auditable.' Reference-construction is the natural concrete demonstration anchor.

3. DEMONSTRATION DESIGN (≤80 words)

Longitudinal deployment of WorkGraph across the founders' active genomics research programs (Erik's pangenomics/reference-building, Luca's clinical genomics analysis). Real workflows, real outputs (preprints, software releases, datasets), all coordinated through auditable computation graphs. Vaughn does ethnographic case studies. BioBench is the open benchmark. Lab adoption is the falsifiable external metric.

4. WHAT TO KEEP FROM V2 (bullet list)

Identify sections that should carry forward verbatim from v2-with-audit-fixes (team bios, budget, COI, contact info, certifications, organizational structure, anything the reframe doesn't touch). Cite section numbers.

5. WHAT TO REPLACE (bullet list)

Identify sections that need full reframing (problem/approach/demonstration/impact/track record framing/theory of change/risks). Cite section numbers.

6. EXPLICITLY DO NOT CLAIM (bullet list)

  • Specific clinical results in 36 months
  • A PHR-into-rare-disease diagnostic pipeline
  • Drug discovery
  • Direct collaboration with Casgevy developer (CRISPRme finding was independent)
  • 'Wrote this proposal with WorkGraph' recursion claim
  • That WorkGraph will produce the next human reference (deployed IN reference-construction workflows is fine; replacing curators is not)

7. LIVERPOOL HIVE MIND POSITIONING (≤60 words)

Per CLAUDE.md: Liverpool's inaugural-cohort Hive Mind grant addresses hybrid coordination for wet-lab autonomous chemistry. Poietic PBC addresses the parallel architectural pattern for computational and clinical genomics. Different domain, different coordination problem, complementary slot. Use this framing in §28. Do NOT do independent web research; CLAUDE.md is sufficient.

8. STYLE INVARIANTS (bullet list)

  • No em-dashes
  • No 'PI' / 'lead PI' language
  • Founder order: Erik, Luca, Vaughn
  • No specific effort percentages
  • No v1 terminology (KRAS, MRTX1133, pancreatic, Boltz, RFdiffusion, DiffDock)
  • CRISPRme/Casgevy framing: Canver 2015 + independent CRISPRme 2023 finding, no implied direct collaboration

Constraints

  • Hard cap: 500 words total.
  • Be terse. The fanout drafters will read this and need to internalize fast.
  • This is a reference doc, not a manifesto. No prose flourishes.
  • Output at ~/poietic.life/notes/v3-spine-brief.md.

Validation

  • All eight sections present
  • PBC benefit statement quoted verbatim somewhere
  • Section numbers cited in 'keep' and 'replace' lists
  • Liverpool framing one short sentence, drawn from CLAUDE.md
  • Total under 500 words
  • Output at ~/poietic.life/notes/v3-spine-brief.md

Depends on

Required by

Log