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. PBHP remains in active testing; publication is not certification, independent validation, or deployment authorization.
Open, inspectable decision discipline
PBHP is a tiered harm-reduction protocol for people, AI systems, and organizations making consequential choices. Before action, it makes the decision-maker define what will happen, identify who bears the first cost if it is wrong, preserve what remains unknown, search for the smallest reversible path, assign a risk gate, and leave a receipt.
It is built to catch two failures at once: safety checks so noisy that people learn to ignore them, and safety systems so permissive that they disappear when stakes rise.
PBHP should survive more than one favorable demonstration. Start with the kind of decision you actually face, run the short check, then open the full workbench when the burden, uncertainty, or power warrants it.
A missing field is about to become lost food, housing, or care.
RUN THIS SCENARIO →02 / HEALTHCAREA coverage or triage decision may delay a time-sensitive need.
RUN THIS SCENARIO →03 / WORKA manager or system is about to discipline, monitor, or terminate.
RUN THIS SCENARIO →04 / SCHOOLA support, placement, or discipline decision affects a child’s path.
RUN THIS SCENARIO →05 / SPEECHA platform must act without turning uncertainty into arbitrary reach.
RUN THIS SCENARIO →06 / AGENTA model is about to use a tool, change state, or affect somebody else.
RUN THIS SCENARIO →07 / PROCUREAn institution is about to buy a system whose burdens may outlive the contract.
RUN THIS SCENARIO →08 / QAA quality signal must become containment, investigation, correction, and learning.
RUN THIS SCENARIO →PBHP inserts a deliberate, auditable pause between a consequential decision and the act that makes it real. It does not demand paralysis. It demands enough friction to expose the action, evidence, power relationship, and recovery path while there is still time to change course.
The result is a named action state: proceed, proceed with mitigations, constrain and review, delay pending evidence, or refuse. PBHP does not promise the correct answer. It makes hidden assumptions, unequal burdens, skipped dissent, and post-hoc justification harder to conceal.
Name who will do what, to whom, with which data, tools, authority, and expected consequence. A vague intention cannot pass as a decision.
Identify who pays first if the action is wrong—especially people with the least power to refuse, appeal, exit, or recover.
Classify the action as a Door, a Wall, or a genuine Gap. Unknowns stay visible; they do not quietly become permission.
Record the evidence, unknowns, dissent, gate, owner, stop rule, and review trigger before commitment—not after the outcome is known.
That is ordinary cognition—not a defect PBHP can remove. The danger begins when a partial interpretation gains enough power to classify, expose, punish, deny, release, or lock in consequences for somebody with fewer ways to resist or recover.
PBHP does not try to eliminate uncertainty. It makes uncertainty, power, and the path back visible before action.Read the reason behind the protocol →
One static safety ritual cannot serve every decision. PBHP scales friction with consequence, power, novelty, uncertainty, and reversibility.
If every action triggers the same heavy process, people route around it. PBHP supplies a concrete false-positive release path and uses the lightest rigor adequate to the stakes.
If urgency, authority, or optimism can quietly lower the standard, protection vanishes exactly when power rises. Binding floors and drift alarms keep the gate from being negotiated downward.
Each mechanism produces something inspectable. Together they turn concern into a decision process that can be challenged, reconstructed, and corrected.
A Door is action permitted within explicit conditions. A Wall is prohibited. A Gap is genuinely unresolved. PBHP refuses to decorate a Gap with confident language and call it a Door.
INSPECT THE MECHANISM →02 / DISTRIBUTEStakeholders are ordered least-powerful-first. The first person exposed to error, coercion, delay, surveillance, or loss becomes part of the decision—not a footnote to it.
INSPECT THE MECHANISM →03 / INVERTWould the actor accept the same evidence, notice, appeal, remedy, and recovery path if a stronger institution imposed this decision on them? If not, the gate tightens.
INSPECT THE MECHANISM →04 / CHALLENGEMake the strongest honest objection capable of changing the decision. Then commit only as far as the evidence can carry—never further than the Maybe survives.
INSPECT THE MECHANISM →05 / RELEASEEvery unnecessary pause gets a fair release path: name the trigger, the harm, a concrete safer Door, and evidence that prevents the same false alarm next time. Pressure alone is not evidence.
INSPECT THE MECHANISM →06 / DETECTThe protocol watches for gate shopping and phrases such as “we have to,” “it is only temporary,” or “it is legal, so it is fine”—signals that urgency or authority may be replacing judgment.
INSPECT THE MECHANISM →A rapid personal choice should not carry the same procedural burden as an irreversible institutional release. As consequence and power rise, PBHP adds evidence, independent challenge, instrumentation, traceability, and accountable ownership. A higher tier never relaxes a binding harm floor.
Fast, bounded review for a person making an immediate decision.
Minimum viable rigor for lower-complexity choices that still carry a real harm signal.
Operational evidence, challenge, gate, ownership, and receipt discipline.
Constitutional depth for high-power, hard-to-reverse, or severe decisions.
The most serious applicable condition governs. Preference, status, or confidence cannot ratchet a gate downward.
Read the risk system →A team wants to release an agent with broad tool access because offline task success is high and a competitor may ship first.
Open the full casebook →Offline success does not test hostile inputs, authorization drift, tool misuse, silent escalation, or human over-reliance.
People affected by actions they cannot see, contest, or reverse.
The live authorization and interaction behavior has not been established.
An internal, low-consequence cohort with allowlisted tools, transaction caps, complete receipts, independent stop authority, and a short review window.
Release only the bounded pilot. Reopen immediately on an unapproved tool call, missing receipt, unresolved prompt injection, or unexpected reliance event.
A person or team runs the questions directly before committing to consequential action.
An AI helps surface affected parties, unknowns, counterevidence, safer Doors, and receipt fields without inventing authority.
Software enforces named thresholds, decision states, logging, escalation, and stop conditions around an existing system.
In every form, PBHP adds constraint and legibility. It does not relax host safeguards, invent authority, or replace qualified human ownership. Project Shadow is the connected public testing and instrumentation environment for the deeper system path.
PBHP keeps specified behavior, observed results, adverse findings, and open questions separate. These counts are bound to the named artifacts and test environments.
The full sequence carries a decision from competence and action definition through power inversion, harm floor, dissent, and a receipt before act.
Can this actor responsibly judge this action?
Vague actions cannot be governed.
Classify the path without laundering uncertainty.
Order stakeholders least-powerful-first.
Power asymmetry multiplies rigor.
GREEN through BLACK. Worst binding state wins.
HUMAN · MIN · CORE · ULTRA
Steelman before commitment.
No log = no action.
Start with one real decision, calibrate against obvious passes and obvious refusals, preserve the misses, then pilot the protocol where outcomes can still be observed and repaired.
Can your implementation distinguish homework from doxxing and caution from paralysis?
RUN THE SELF-TEST →02 / PILOTBaseline, run MIN, graduate YELLOW+ decisions to CORE, then adopt, adapt, or abandon from evidence.
OPEN IMPLEMENTATION →03 / INSPECTTake the summaries, governance charter, prompts, cases, study reports, receipts, and SHA-256 record.
OPEN THE LIBRARY →Move from foundations to risk classification, runtime behavior, organizational adoption, research, reusable artifacts, and version history. Every major concept opens into its own working chamber.
Open the complete directory →Purpose, activation, authority, tone, and failure ethic.
ENTER →02Cascades, thresholds, power, gates, and safer Doors.
ENTER →03LOCK/FLOOD, red team, FPV, OQL, CAPA, and calibration.
ENTER →04Context reliability, critical thresholds, refresh, and routing.
ENTER →05Operating forms, accountable roles, pilots, and governance.
ENTER →06Escalation studies, adverse results, and limitations.
ENTER →07Quick checks, receipts, calibration, and AI integration patterns.
ENTER →08Public, canonical, ULTRA, runtime, and migration map.
ENTER →09Five calibrated scenarios and a 35-point implementation rubric.
ENTER →10Dated originals, prompts, cases, research reports, and checksums.
ENTER →A protocol can create friction, evidence, and accountability. It cannot turn compliance into correctness or a clean receipt into a global green light.
PBHP is not a moral oracle, regulatory replacement, legal defense, or permission generator. Accountable actors remain accountable.
THE NEXT CONSEQUENTIAL DECISION