INSTITUTION
An agency, legislature, court, school, platform, workplace, public authority, or contractor.
PROJECT SHADOW 1.0.1 · CORRECTED R1 REFERENCE · PRELIVE · 2026-08-17
Project Shadow 1.0.1 contains no Myth package. Generic Myth v0.2.0 and Full-Canon Myth v0.3.5 are separate optional companions; both default off, neither is required by R1, and neither can authorize action or change an R1 result. Civic QA remains self-service and unofficial; publication transfers no authority and is not operational deployment, efficacy, safety, certification, a legal-compliance finding, permission to act, or a global green.
ECOSYSTEMCIVIC QA
One civic lifecycle / six distinct public doors
Project Shadowinstruments + testsPBHPpausesCivic QAinspectsARMrepairsThe RecordremembersALMSIVIpreserves optional originInspection before trust
INSPECT PUBLIC
SYSTEMS BEFORE
TRUSTING THE CLAIM.
Civic QA gives anyone a transparent, repeatable way to define a civic claim, test it against evidence, expose hard stops and unknowns, build corrective action, and export a challengeable receipt.
ANYONE CAN RUN + EXPORT · SELF-SERVICE IS NOT OFFICIAL · ONLY A REVIEWED EDITION CARRIES THE CIVIC QA MARK
CIVICQAQUALITY ASSURANCE FOR PUBLIC CONSEQUENCE
- 01Authority
- 02Evidence
- 03Burden
- 04Rights
- 05Capacity
- 06Data + AI
- 07Measurement
- 08Correction
REQUIREDCLAIM · SOURCE · FINDING · OWNER · CAPA · EFFECTIVENESS TEST
THE DIFFERENCE BETWEEN PAPER AND REALITY
Closed is a paperwork state. Fixed is an evidence state.
A policy can be signed, a training assigned, a dashboard turned green, and a corrective action closed while the burden remains exactly where it was. Civic QA keeps the finding open until the repair survives an effectiveness check in the world.
Eight workbench screens. One local receipt.
No account or special credential is required. These are the interface screens—Choose through Export—not the separate eight-phase public inspection lifecycle below. The workbench stays local, recomputes every gate, and never grants itself an official mark.
- 01
Choose
Select legislation, candidate accountability, local service, health/benefit, AI procurement, or agency performance.
- 02
Define
Name the exact claim, authority, version, time window, affected people, geography, purpose, and exclusions.
- 03
Source
Build an attributable ledger and keep fact, verification, inference, hypothesis, and unknown apart.
- 04
Inspect
Answer eight shared dimensions and the profile-specific safeguards with rationale and evidence IDs.
- 05
Gate
Let scope, hard stops, material unknowns, and major findings govern before the total.
- 06
Find
Write the condition, criterion, consequence, recurrence mechanism, owner, due date, and closure evidence.
- 07
Repair
Define correction, prevention, verification, threshold, failure trigger, and ARM handoff.
- 08
Export
Copy or download Markdown/JSON, print a review packet, and preserve the correction path.
Any system that converts public authority into consequence.
The Record supplies evidence and memory. Civic QA tests the mechanism against explicit standards. The American Repair Manual carries the finding into corrective and preventive action.
POLICY
A law, rule, executive action, budget, program, procurement, enforcement practice, or implementation plan.
PROMISE
A candidate, party, official, vendor, or reform proposal claiming public benefit or accountability.
AI SYSTEM
A model, ranking system, decision aid, monitor, assistant, or agent crossing into institutional consequence.
TWELVE APPLICATION PATTERNSDomain guidance for Congress, candidates, agencies, healthcare, schools, courts, procurement, AI, media, and more. The workbench implements six broader operational profiles.OPEN THE APPLICATION MAP →
Evidence in. Inspection through. Receipt out.
These governance handoffs run Define through Publish. They are distinct from the workbench's eight navigation screens. Each phase leaves a documented handoff and can pause urgent harm without pretending the larger system is repaired.
- 01
Define
Name the institution, decision, policy, system boundary, time period, authority, and people affected.
- 02
Source
Open a ledger. Separate observed fact, external verification, inference, hypothesis, and unknown.
- 03
Inspect
Run explicit standards against design, implementation, distribution, rights, evidence, AI use, and control.
- 04
Find
Write the deviation, evidence, impact, severity, owner, and the condition required for closure.
- 05
Pause
If harm is live, preserve the smallest safer reversible path while the larger repair remains open.
- 06
Repair
Correct the present failure and change the mechanism that would otherwise reproduce it.
- 07
Verify
Test effectiveness against a prewritten threshold, including burden distribution and unintended effects.
- 08
Publish
Release the finding, source coverage, dissent, correction history, result, and unresolved questions.
Look past the policy text. Test the operating system.
Every dimension asks for evidence and a consequence path. An attractive average never overrides a critical rights, authority, recourse, or AI control failure.
Authority & mandate
Is the power being exercised lawful, explicit, bounded, visible, and matched to the stated purpose?
01 / 08Evidence & claim integrity
Are decision-bearing claims sourced, current, reproducible, uncertainty-labeled, and open to counterevidence?
02 / 08Access & burden
Who waits, pays, proves, travels, discloses, complies, or loses access—and can they realistically complete the process?
03 / 08Rights & recourse
Are notice, explanation, privacy, due process, appeal, remedy, and recovery usable by the people most affected?
04 / 08Delivery & capacity
Do staffing, funding, procurement, training, interfaces, and operational dependencies support the promised result?
05 / 08Data, AI & automation
Are data use, objective, model limits, permissions, testing, human ownership, monitoring, stop authority, and retirement conditions visible?
06 / 08Measurement & distribution
Are success, harm, quality, access, and unintended effects measured across groups without hiding people in aggregate averages?
07 / 08Correction & public accountability
Can evidence change the system, is recurrence prevented, and does the public receive a complete record even when the result is poor?
08 / 08WORST MATERIAL FINDING GOVERNSNo global green. Publish the scores, the hard stops, the evidence coverage, and the unknowns.OPEN SCORECARD →
Dated public questions, with the evidence limits left in.
Congressional stock trading. Shreve and Wirth. RequestIndy closure. Indiana Medicaid prior authorization. Benefits AI procurement. SSA disability backlogs. Each case retains its stated evidence cutoff and remains Phillip-review pending; none is represented as current today.
OPEN THE SIX-CASE LEDGER →- CONGRESS
Disclosure controls exist; a blanket stock-trading ban is not enacted.
- IN-6
Ask identical questions and publish an evidence inventory—not a winner.
- INDIANAPOLIS
“Closed” is not the same evidence state as resident-confirmed resolution.
- HEALTH + AI
Documented timelines, awards, and policies do not by themselves validate access or safe use.
- AGENCY CLAIM
SSA disability performance improved; a material backlog and evidence gaps remain. Improved is not resolved.
OFFICIAL MARK / GOVERNED PUBLICATION
Official does not mean infallible.
It means one exact edition passed review for scope, source traceability, rationale, hard-stop handling, affected-party burden, correction path, and disclosure. Material new evidence reopens it. A correction preserves the prior edition. A withdrawal leaves a public tombstone.
START WITH ONE REAL SYSTEM