Runwisp / proof acceptance reference

A reviewer's reference for laboratory proof runs.

Use this page to verify the preview behavior around normalization, exceptions, redacted examples, and human review. It is an acceptance reference for a proof run—not a compliance guarantee, validation decision, or substitute for the laboratory's existing tools.

Static reference

Synthetic examples

No live run data

Proof boundaries

What this reference does—and does not—establish.

Data: synthetic or fully redacted only.

Purpose: a QC and reporting preview for review.

Workflow: Runwisp stays alongside existing laboratory tools.

01 / Normalization accuracy

Make the source legible without hiding what changed.

Reviewers should be able to compare the source shape with the canonical fields, the normalized value, and the reason a row was accepted or held for review.

Header normalization

Remove a leading BOM, match headers without case or punctuation sensitivity, and map aliases to the canonical fields method, instrument, batch, control, result, status, and optional reviewer notes.

Result normalization

Accept numeric forms such as 0.004, <0.001, and 4.2e-3. Preserve a qualifier such as “<” for review context; flag values that cannot be interpreted as numeric results.

Date normalization

Accept YYYY-MM-DD and MM/DD/YYYY, then emit the normalized date expectation. Flag impossible dates or dates in an unsupported format.

QC flag normalization

Map pass, passed, ok, accepted, in limits, yes, and true to pass. Map fail, failed, rejected, out of limits, no, and false to fail. Flag every other value.

Supported profile expectations

Use the profile that matches the export shape; do not force an unsupported layout into a passing result.

Required fields by proof profile.
Profile	Required expectations	Canonical context
Named-method export	Method, instrument, batch, measurement, units, status	Optional reviewer notes can travel with the row.
Wide-instrument export	Sample ID, analyte, result, units, analysis date, QC flag	Instrument exports must also satisfy the recency check.

Count reconciliation

Reviewers verify total rows = valid rows + invalid rows. The visible proof preview is limited to 8 rows; exception output is limited to 20 entries.

Status is not inference

Numeric parsing and alias mapping make a value legible; they do not decide whether the laboratory result is scientifically valid or compliant.

02 / Error flags

Every exception should tell a reviewer what to inspect next.

A row-level issue identifies the field, gives a human-readable message, and includes a fix hint. When a validation error exists, the row is not accepted. Top-level exceptions are grouped so qualification failures are not mistaken for ordinary row cleanup.

Redaction

Sensitive-looking headers or values, or missing proof of redaction.

Qualification

Attestation or recency conditions that prevent the run from qualifying.

Format

Malformed CSV, ambiguous headers, unsupported layout, or extra values.

Validation

A required value, result, date, or QC flag fails row-level validation.

Recency

A wide-instrument export is older than the 90-day acceptance threshold.

Expected outcomes matrix

Every negative condition has a visible category and a review outcome.

Acceptance behavior for common proof-run exceptions.
Condition	Expected flag or action	Outcome
Required value is blank	Flag the field as validation; do not accept the row.	Review
Header is missing, duplicated, or maps to more than one canonical field	Flag the mapping as format; do not qualify the run.	Review
Columns do not match a named-method or wide-instrument profile	Flag unsupported layout as format; request a corrected export.	Review
Malformed CSV quoting, delimiters, or line structure	Flag malformed CSV as format; no row-level acceptance.	Review
A row contains more values than the header defines	Flag extra row values as format; rerun after source correction.	Review
Result is not numeric, such as “trace” or “n/a”	Flag result as validation, retain the original value for review, and reject the row; proof qualification is evaluated separately.	Review
Date is impossible or unsupported, such as 02/31/2026	Flag analysis date as validation, emit no normalized date, and reject the row; proof qualification is evaluated separately.	Review
QC flag is unknown, such as “maybe”	Flag QC flag as validation, do not infer pass or fail, and reject the row; proof qualification is evaluated separately.	Review
The file parses but contains zero data rows	Flag zero data rows as format; the run is not qualified.	Review
Wide-instrument export is more than 90 days old	Flag recency; human review is required even if rows normalize.	Review

Qualification boundary: a row-level validation exception is different from proof qualification. Missing attestation, redaction, qualification, format, or recency exceptions prevent qualification even when some rows normalize successfully.

03 / Representative redacted inputs

Use fixtures that prove the path without exposing laboratory data.

These fixtures are synthetic. They contain no names, emails, phone numbers, accession numbers, secrets, or other direct identifiers. A reviewer should be able to paste the shape into a test harness without importing a real record.

Wide-instrument CSV / synthetic pass

Sample ID,Analyte,Result,Units,Analysis Date,QC Flag
LAB-001,AN-204,0.004,mg/L,2026-08-14,pass
LAB-002,AN-204,<0.001,mg/L,08/15/2026,accepted
LAB-003,AN-204,4.2e-3,mg/L,2026-08-16,ok

Named-method CSV / one review flag

Method,Instrument,Batch,Measurement,Units,Status,Reviewer Notes
EPA-204.2,INST-SYN-01,BATCH-26-001,0.004,mg/L,passed,synthetic control
EPA-204.2,INST-SYN-01,BATCH-26-001,0.006,mg/L,failed,inspect control context

Example 01

Successful normalization

Accepted / qualified

Aliases resolve to canonical fields and each synthetic row is accepted.

Redacted input / CSV

Method,Instrument,Batch,Control,Result,Analysis Date,Status
M-204,INST-SYN-01,BATCH-26-001,LAB-001,0.004,2026-08-14,pass
M-204,INST-SYN-01,BATCH-26-001,AN-204,<0.001,08/15/2026,accepted
M-204,INST-SYN-01,BATCH-26-001,AN-204,4.2e-3,2026-08-16,ok

Expected normalized result

valid: true
rowsAccepted: 3
reviewFlags: []
method: M-204
instrument: INST-SYN-01
batch: BATCH-26-001
control: LAB-001 | AN-204 | AN-204
result: 0.004 | <0.001 | 4.2e-3
date: 2026-08-14 | 2026-08-15 | 2026-08-16
status: pass | pass | pass

Reviewer outcome

Confirm the canonical dates preserve the source values and the “<” qualifier remains visible. No exception is emitted.

Example 02

Flagged row-level error

Row rejected / run assessed separately

A concrete bad value is retained for diagnosis, but the row is not accepted.

Redacted input / CSV

Method,Instrument,Batch,Result,Analysis Date,Status
M-204,INST-SYN-01,BATCH-26-001,trace,2026-08-14,pass

Expected flagged result

valid: false
validation: result must be numeric
method: M-204
instrument: INST-SYN-01
batch: BATCH-26-001
original.result: trace
normalized.result: omitted
normalized.status: pass
rowAccepted: false

Reviewer outcome

Correct the source value or document the source context. Other valid rows may still be counted; a row error does not by itself claim the entire proof run is blocked.

Example 03

Valid row escalated to human review

Row accepted / qualification held

The row is structurally valid, but a failed QC status requires source and control context.

Redacted input / CSV

Method,Instrument,Batch,Result,Analysis Date,Status
M-204,INST-SYN-01,BATCH-26-001,0.006,2026-08-14,rejected

Expected normalized result

valid: true
method: M-204
instrument: INST-SYN-01
batch: BATCH-26-001
result: 0.006
date: 2026-08-14
normalized.status: fail
reviewFlags: [qc-fail]
rowAccepted: true
qualification: held for human review; not blocked by row validation

Reviewer outcome

Inspect the control record and source context, then decide whether the proof remains qualified. Normalization does not convert fail or rejected into pass.

Compact negative fixtures

Each fixture states the exact expected flag/category and separates row acceptance from proof qualification.

Synthetic negative cases only.
Fixture	Input	Expected flag/category	Proof-level outcome
Blank required value	LAB-001,AN-204,,0.004,2026-08-14,pass	validation / batch is required	Row rejected; proof qualification evaluated separately
Non-numeric result	LAB-001,AN-204,BATCH-26-001,trace,2026-08-14,pass	validation / result must be numeric	Row rejected; proof qualification evaluated separately
Invalid date	LAB-001,AN-204,BATCH-26-001,0.004,02/31/2026,pass	validation / analysis date is invalid	Row rejected; proof qualification evaluated separately
Unknown QC flag	LAB-001,AN-204,BATCH-26-001,0.004,2026-08-14,maybe	validation / QC flag is unknown	Row rejected; proof qualification evaluated separately
Duplicate header	Sample ID,Sample ID,Analyte,Result,Units,Analysis Date,QC Flag	format / ambiguous sample ID mapping	Run blocked; corrected mapping required
Sensitive-looking header/value	Patient Name,AN-204,BATCH-26-001,0.004,2026-08-14,pass	redaction / prohibited sensitive field detected	Run blocked; redaction required

04 / Human-review triggers

A preview is useful when it makes the next human action obvious.

Review is required whenever the source, mapping, attestation, or result needs a person to resolve context. The expected action is to inspect the source row and surrounding context, correct or explain the input, rerun the proof, and never treat the preview as automatically accepted or as a compliance determination.

Row-level validation errors

Inspect the source row and correct or explain the input before rerunning.

Canonical QC status is fail

Treat the result as a human-review flag, never as an automatic acceptance.

Redaction finding or missing attestation

Stop qualification until the source is redacted and the attestation is present.

Missing, ambiguous, or unsupported mapping

Confirm the intended profile and correct the export or mapping context.

Malformed CSV or zero data rows

Repair the source file and rerun the proof; do not interpret an empty preview.

Missing recent-export attestation

Request the attestation before treating the proof as qualified.

Wide-instrument export older than 90 days

Review the source date and confirm whether a newer export is available.

Inspect

Check the source row, headers, and surrounding context.

Correct or explain

Resolve the input or document why the source is intentionally different.

Rerun

Run the proof again; do not promote a held preview to acceptance.

05 / Measurable acceptance checklist

Done means a reviewer can verify the behavior from the page alone.

Use this checklist at the end of a proof review. It intentionally measures what is visible to a reviewer instead of requiring implementation context.

Findable topics

A reviewer can locate normalization accuracy, error flags, redacted examples, and human-review triggers without reading source code.

Visible outcomes

Every listed fixture shows a normalized value or exact flag/category and a pass or review outcome.

Measurable limits

The page states total rows = valid rows + invalid rows, an 8-row preview cap, a 20-exception cap, and a 90-day recency threshold.

Safe examples

Examples use synthetic or redacted data only and explicitly prohibit names, emails, phone numbers, accession numbers, and secrets.

Reachable and readable

The route is linked from the public footer, has canonical metadata, and remains readable on narrow screens with no API, auth, database, or module dependency.

Reference outcome

A proof run is review-ready only when the visible counts, flags, fixtures, attestations, and recency checks agree.

That agreement supports a focused human review. It does not make Runwisp the laboratory's validation authority or replace the tools and procedures already in use.

06 / Proof-run completion standard

Completion is an evidence gate, not a status shortcut.

The admin handoff uses the same requirements below before it accepts a completed disposition. A reviewed exception with a required source correction remains in review until the correction is explicitly resolved.

Safe redaction confirmed

The proof uses synthetic or fully redacted data and the operator attested to it.

Recent export attested

The source export is confirmed to be recent enough for this proof review.

Supported file layout

The stored source profile matches a supported proof import layout.

At least one valid result

The stored preview contains a non-empty set of valid result rows.

Validation state is clear

No invalid rows or rejected/exception validation state remains on the run.

Every exception reviewed

Each stored exception has an operator disposition and required review note.

Required fixes resolved

An exception with a source correction hint is resolved, not merely reviewed.

Readiness evidence recorded

The linked pilot request has a durable evidence reference for readiness.

Transition readiness is ready

The linked pilot request is explicitly marked ready for transition.

Primary contact role verified

The operator has verified the primary contact role.

No proof-run exit blockers

The existing exit projection has no stop indicator or unresolved blocker.