Novel Biotech — current progress

Concept study · Research only · No human testing · Not available · No development program exists

The accepted record ends at NO-SELECTION. Candidate evidence is UNKNOWN, comparator evidence is UNKNOWN, and the exact complete-match search is OPEN. This is a hold—not evidence failure and not permission to proceed.

Verdict REVISIONDecision HOLD_BOTH_UNKNOWNRelease Unauthorized

Executive Summary

No target is selected. Step 5 accepts a target-adjudication record whose deterministic result is HOLD_BOTH_UNKNOWN.
S5R2-T01 is deferred, not chosen. It concerns vision-occluded contact/release object manipulation in adults with acquired unilateral transradial limb loss: direct natural peripheral ENG forward, afferent peripheral-nerve stimulation return, and an optimized removable surface-EMG plus external vibrotactile or electrocutaneous comparator.
Both complete routes remain unknown. Component studies cannot be pooled into a complete candidate-versus-comparator result. T02 is DEFER; T03, T04, and T05 are HOLD, not evidence FAIL.
The work remains research control. Families A/B/C are unselected; Legal/Regulatory/Ethics is an independent gate, not clearance; R0 is blocked; R1 bench v1 is rejected; release remains unauthorized.

The Few Terms Needed

ENG: electrical activity recorded directly from a peripheral nerve. It is not muscle activity, an evoked response, or stimulation alone.
EMG: electrical activity recorded from muscle. Surface, implanted, or RPNI-derived EMG remains muscle-derived and cannot be relabelled as direct ENG.
HOLD_BOTH_UNKNOWN: candidate and matched comparator both lack eligible complete-route evidence. The rule holds both; it does not infer failure or grant permission.
DEFER / HOLD / KEEP AS RESEARCH BRANCH: DEFER waits behind prerequisites; HOLD means the decision cannot close; KEEP preserves comparator research without selection.
Accepted record / landed in git: accepted bytes passed review for one narrow purpose. Landed bytes entered local git history. Neither state authorizes implementation or release.

Where We Are Now

BPNI bounded the branches and comparators. The Limen parent enforced route identity. Steps 3 and 4 closed communication and decision-input records. Step 5 applied the frozen rule and returned NO-SELECTION.

Step 3 is concept-architecture and blueprint communication, not a physical device. Step 4 is input-record closure, not architecture selection.

What Is Accepted—and What Is Not

Accepted purpose and forbidden implication stay separate.

Current workstream status
Workstream	Accepted as	Not authorized
BPNI	Research-control bytes accepted 12 Aug 2026. Motor loss remains a comparator-narrowing research branch; sensory, severe-paralysis access, autonomic/organ, pain, and rehab are deferred. Diagnostics/monitoring is rejected for now with KILL-04 unfired. B00 was N=2 over 4–5 weeks with mixed neural-plus-EMG control; in B01, 6 of 7 arrays lost functional recording electrodes within about two months. Science hashes match but remain outside local git because standalone repository-QA evidence is missing.	Target selection, durable direct-ENG inference, simultaneous-duplex inference from same-person recording plus stimulation, or substitution of muscle EMG, RPNI EMG, evoked CAP, or stimulation-only return.
Limen Step 3	Concept-architecture and blueprint communication.	A physical device, component selection, build specification, implant, or development program.
Limen Step 4	Input-record closure and deterministic decision rule.	Architecture A/B/C, device, indication, biological interface, or implementation selection.
Limen Step 5	Explanatory and target-adjudication record accepted 23 Aug 2026: NO-SELECTION. S5R2-T01 candidate UNKNOWN; comparator UNKNOWN; exact complete-match search OPEN; result HOLD_BOTH_UNKNOWN.	Converting S5R2-T01 into an authorized effort, pooling component studies, or treating missing evidence as failure or permission.
R0	No accepted implementation; only a possible synthetic, offline, non-actuating software stage.	Starting before target selection, the ten technology decisions, and Science-accepted synthetic reference vectors.
R1 bench v1	Rejected and ineligible.	Reuse, purchase, assembly, testing, or any implication that rejection was reversed.
Website	Graphite work exists locally.	Intake, deployment, publication, or implication of an operating company, laboratory, product, or program.
Landing queue	Tracker cutover, Limen Science parent, Blueprint Inputs, and Step 4 Science child landed locally in commit `52657fc`. Product Engineering R3 is next. Blocked behind it: Step 3 Product, Step 4 remaining lanes, then Step 5 Product, Science, Legal, and founder packages.	Push, skipped review order, or claims of feasibility, safety, efficacy, novelty, or availability from acceptance or landing.

What Would Change the Step 5 Decision

Only eligible complete-route evidence under the frozen rule can change the result. The candidate and comparator must be tested in the same population, against the same function and endpoint, across the same observation window.

Candidate route: direct natural peripheral ENG forward plus afferent peripheral-nerve stimulation return, with burden and adverse events explicit.
Matched test: population, function, endpoint, duration, denominator, interruptions, revisions, explants, category identity, and recalibration burden.
Comparator route: optimized removable surface-EMG plus optimized external vibrotactile or electrocutaneous contact/release feedback.

Current output: HOLD_BOTH_UNKNOWN. Candidate UNKNOWN plus comparator UNKNOWN stays on hold. Another bounded negative search leaves the exact complete-match search OPEN.

PMID 33725672 is N=1 across three recording days and reports similar TIME neural versus surface-EMG force/velocity decoding. Unique direct-neural advantage fails only at that narrow forward-only endpoint; there is no return route, adequate chronic window, or complete-route result. PMID 26584497 is N=5 for one month of external contact/release feedback. It raises the comparator bar but is not an eligible complete-route head-to-head. Component studies cannot be pooled.

What to Do Next

Keep the exact complete-match search open. Do not broaden eligibility or pool component studies.
Extract both routes symmetrically. Record route, population, endpoint, duration, denominator, adverse events, revision, explant, interruption, category identity, and recalibration burden.
Reapply the frozen table. UNKNOWN plus UNKNOWN remains HOLD_BOTH_UNKNOWN.
Commission independent review on one frozen record. Include affected-community, clinical, disability-rights, bioethics, privacy, cybersecurity, regulatory, and jurisdiction-specific review.
Translate only after separate target acceptance. Product may then translate neutrally and Backend may resolve the ten R0 technology decisions.
Accept exact synthetic vectors separately. Science acceptance is required before any software-only, non-actuating R0 simulator.
Keep R1 downstream. Any non-biological bench reconsideration requires accepted R0 evidence and explicit founder authorization. Rejected R1 bench v1 remains ineligible.

Report or record	Why open it
Library index	Full report map.
Founder-report READMEPrivate repository source	Sibling-report custody and scope.
Current Step 5 HTML	Accepted explanation of NO-SELECTION.
Step 5 decision in prosePrivate repository source	Controlling logic without presentation.
Step 5 glossaryPrivate repository source	Routes, states, comparators, and evidence terms.
Step 3 founder brief	Concept communication and non-device boundary.
Step 4 founder brief	Input closure and frozen hold rule.
Science founder brief	Older snapshot; not the Step 5 decision.
Duplex contract	Why same-person components do not establish simultaneous duplex.
BPNI research-round PDFPrivate repository source	Accepted research-control baseline.
Step 5 Science READMEPrivate repository source	Scope, inventory, and decision boundaries.
BPNI baselinePrivate repository source	Pre-Limen indication and comparator dispositions.
HANDOFFPrivate repository source	Canonical integration state and next gates.
PRODUCTPrivate repository source	Public truth and claim boundaries.
Mixed-origin Step 5 candidateDo not use	Historical bytes excluded from authority.
Unaccepted Step 5 predecessorDo not use	Superseded explanatory report.
Rejected R1 bench v1Private repository sourceDo not use	Rejected; no build, purchase, assembly, or test authority.