pub struct RunConfig {Show 13 fields
pub tools: Option<Vec<ToolSchema>>,
pub available_skills: Option<Vec<SkillMetadata>>,
pub stable_core_tools: Option<Vec<String>>,
pub memory_enabled: Option<bool>,
pub knowledge_enabled: Option<bool>,
pub plan_tool_enabled: Option<bool>,
pub tokenizer: Option<String>,
pub governance: Option<GovernanceConfig>,
pub attention_max_queue_size: Option<u32>,
pub scheduler_max_wall_ms: Option<u64>,
pub resource_quota: Option<ResourceQuota>,
pub group_tokens_base: Option<u64>,
pub group_spawns_base: Option<u32>,
}Expand description
K2: a bundle of run-setup configuration carried by the KernelInputEvent::ConfigureRun event.
Each field maps 1:1 to a granular Set* / Load* event; None/absent leaves that aspect untouched.
This is the host-side analogue of the SDK’s applyKernelPolicies — one event for the whole setup.
Fields§
§tools: Option<Vec<ToolSchema>>§available_skills: Option<Vec<SkillMetadata>>§stable_core_tools: Option<Vec<String>>§memory_enabled: Option<bool>§knowledge_enabled: Option<bool>§plan_tool_enabled: Option<bool>§tokenizer: Option<String>Present (any value) ⇒ reset the token engine to the char-approx estimator (see SetTokenizer).
governance: Option<GovernanceConfig>§attention_max_queue_size: Option<u32>§scheduler_max_wall_ms: Option<u64>§resource_quota: Option<ResourceQuota>§group_tokens_base: Option<u64>L1 (RunGroup): cumulative tokens already spent by other members of this run’s governance
domain, seeded at boot so the run-level token cap (max_total_tokens) is enforced across the
whole group, not per-vehicle. None/0 ⇒ no group (N=1) ⇒ pre-L1 per-kernel behavior.
group_spawns_base: Option<u32>L1 (RunGroup): sub-agents already spawned by other members of this run’s governance domain,
seeded at boot so ResourceQuota::max_total_subagents is enforced across the whole group.
None/0 ⇒ no group (N=1) ⇒ pre-L1 per-vehicle behavior.