token-budgets-are-accurate-bidirectionally
OUT derived (depth 2)
Token budget management is accurate in both directions: the compact module reliably constrains output size for context-limited consumers, while the derive pipeline correctly allocates input budgets per agent — ensuring resource-bounded operation across the entire LLM integration surface.
Summary
The system's token budget management works correctly end-to-end, meaning both sides of LLM resource control are reliable: output size stays within limits when compacting text, and input budgets are divided accurately when distributing work across agents. This matters because it guarantees the system won't silently overshoot resource limits in either direction. However, this claim is currently marked as unsupported because one or both of its underlying assumptions about compact output sizing and derive budget allocation have been retracted.
Justifications
SL — Both token budget mechanisms (output-constraining compact and input-allocating derive) are independently accurate, yielding bidirectional budget correctness across the LLM boundary
Antecedents (all must be IN):
- compact-budget-controls-output-size — The compact module's token budget reliably constrains total output size
- derive-budget-allocation-is-accurate — The derive pipeline's proportional belief-budget allocation produces correct per-agent token counts
Dependents
These beliefs depend on this one:
- information-pipeline-is-resource-governed-and-access-controlled — The complete information pipeline is governed along two orthogonal axes: token budgets accurately constrain both input (proportional derive allocation) and output (compact distillation with budget enforcement), while access tags enforce transitive subset-based authorization at every read boundary — every piece of information is simultaneously resource-bounded and access-controlled.
- resource-management-supports-belief-currency — Active belief currency management — sustainable derivation of new beliefs and staleness detection for existing ones — operates with accurate bidirectional token budget control, ensuring derivation rounds allocate resources correctly per agent and output fits context-limited consumer constraints.