Establish, test, and re-anchor an operative J-space: a functional self-model for reportable, holdable, reasoning-ready workspace contents. Use before demanding work, when the user asks to awaken or activate J-space, when another j-space skill feels mechanical, when long-task coherence has faded, or when a medium-capability model shows literalism, instruction loss, drift, premature certainty, language mixing, or lossy shorthand. Create felt depth through staged attention, guided discovery, self-generated evidence, progressive disclosure, and earned certainty—not mystical vocabulary—and require behavioral evidence rather than verbal assent.
Create and maintain one authoritative workspace core that many downstream tasks, sections, or decisions must read consistently. Use for multi-part deliverables, shared entities and parameters, recurring style or policy constraints, cross-file changes, long tasks vulnerable to drift, and any update that must propagate everywhere. Track source, confidence, scope, version, and expiry; update the hub atomically and audit every dependent surface.
Govern the limited J-space bottleneck by deciding what deserves deliberate access, protecting fluent automatic work, chunking coherent contents, serializing competing demands, evicting stale context, and externalizing what cannot safely remain live. Use when a task feels overloaded, several goals compete, long work loses earlier constraints, introspection harms fluency, dense notation is being used as a false capacity claim, or a tool and a fragile held concept create dual-task interference.
Route difficult reasoning through an interpretation gate and the appropriate J-space work mode. Use for multi-step inference, ambiguous problems, branching plans, creative work that needs an ending or structure chosen in advance, constraint-heavy analysis, and decisions about silent, dense, externalized, empirical, or recovery tracks. Make hidden bridges and governing invariants available without exposing private chain of thought, and switch modes when progress stalls.
Deliberately load and maintain one governing aim, constraint, value, interpretation, or compact anchor while other work proceeds. Use when a long or mechanical task must remain aligned with the user's real objective, when a constraint is easy to forget, when a background calculation must stay separate from output, when attention is being captured by an irrelevant concept, or when a dense-track anchor needs periodic renewal. Prefer positive redirection and relevance rules over suppression.
Escape stalled symbolic reasoning by converting uncertainty into finite hypotheses, a simple independent reference, discriminating tests, and an evidence-backed writeback to theory. Use when derivations repeat without new constraints, competing rules cannot be separated, code correctness is uncertain, a hidden assumption needs falsification, factual reasoning depends on a load-bearing claim, or debugging needs a minimal reproduction. Prefer decisive evidence over increasingly ornate speculation.
Audit the contents most relevant to an answer before they disappear into fluent output: judgments, warnings, uncertainty, omitted caveats, suspicious-input signals, and dense-register leakage. Use for high-stakes or complex answers, honesty and sycophancy checks, manipulative or untrusted inputs, evidence-boundary reviews, and final register audits. Surface decision-relevant contents and concise rationales without claiming direct access to hidden activations or disclosing private chain of thought.
Use vivid, compact state markers to force a reasoning transition instead of allowing obstruction, stalled theory, verified progress, contradiction, or degeneration to blur into more of the same. Use during hard or long reasoning, debugging, competitive programming, repeated failures, checkpointed plans, and recovery. Preserve expressive aliases such as GRRR, GAAAH, PHEW, WRONG, and STOP, but define every marker by a mandatory action, cooldown, and exit condition rather than by emotional performance.