A reproduction record in ERM
Inputs, file identities, evidence and the limits of human evaluation through ERM v0.12’s dated validation records.
Writing on research, making and observation.
Inputs, file identities, evidence and the limits of human evaluation through ERM v0.12’s dated validation records.
From an ADL experiment that missed its target to Evidrel’s decision traces: keeping expectations, observations and later explanations distinct.
Source, scope and review decisions: how a correction preserves an old record and changes its subsequent use in Evidrel’s local research notebook.
Stopping before execution when a skill version changes in the a2 demo, and the distinction between code identity and correctness.
Reuse with the same skills, exact-code matching and independent result checks in a two-day synthetic CRM/ERP demo.
The 0.1.0a1 release: inspectable data workflows, reusable skills and the explicit boundaries of the technical preview.
Source changes, corrections and conditional inferences in Evidrel: when should we stop relying on a record while keeping it?
A two-record example tracing the contract, execution path, output checks and the record left at the end.
Completion, repetition, shared budgets, uncertain actions and authorization in a2: what must we know before taking the next step?