PATCH semantics. Dashboard-only. Requires member+.
Documentation Index
Fetch the complete documentation index at: https://docs.trulayer.ai/llms.txt
Use this file to discover all available pages before exploring further.
SDK API key (tl_...) or Clerk session JWT
PATCH body — all fields optional; only supplied fields are changed.
eval_score, anomaly, failure_rate retry, fallback_model, route_pct, block, escalate 0 <= x <= 100draft, active, paused, archived Per-policy retry depth cap (TRU-362). Out-of-range values return HTTP 422.
1 <= x <= 10Per-policy cascade depth cap (TRU-371). Counts retry + fallback_model + prompt_modification combined per trace, across policies. Out-of-range values return HTTP 422.
1 <= x <= 20Updated policy
draft, active, paused, archived Raw YAML DSL as authored. Opaque to the server in Phase 1.
Denormalised summary of the when: clause for fast filtering.
eval_score, anomaly, failure_rate Denormalised summary of the then: clause for fast filtering.
retry, fallback_model, route_pct, block, escalate Staged-rollout percentage. 0 means the policy is authored but inactive.
0 <= x <= 100When true, the evaluator records observations without applying the action.
Clerk user ID of the author.
Per-policy retry depth cap (TRU-362). When the executor counts
this many or more previously-executed retry actions on a trace
for this policy, the next retry is auto-converted to an
escalate action and routed through the HITL pending_approval
queue. Default 3.
1 <= x <= 10Per-policy cascade depth cap (TRU-371). Counts every
remediation action on the trace — retry plus
fallback_model plus prompt_modification — across all
policies. When the count reaches this many, the next
remediation is auto-converted to escalate and parked in
the HITL queue. Distinct from max_retry_depth which
counts retries only; both gates run per Execute call and
whichever fires first wins. Default 5.
1 <= x <= 20