The Handoff Said Done. The Repository Did Not.
A reconstructed failure reflection on why AI-assisted work needs exact source identity, bounded verification, and evidence that matches each transition.
// FIELD TRANSMISSIONS
System Logs
An archive of decisions, failures, changes, and shipped work. The record stays useful by keeping the sequence—and the limits—visible.
33entries on record
Filter by domain
Model Choice Became Runtime Policy
Why dynamic model routing turns model selection into execution policy, and what evidence, state, fallback, and evaluation controls it needs before governing real work.
The Loop and the Product Needed Separate Owners
Why agent architecture becomes easier to replace, debug, and govern when reusable execution and host-application authority have separate owners.
The Orchestration Shrunk. The Scaffolding Stayed.
A model-churn test for agent architecture: micro-orchestration may decay as models improve, while context structure, tools, boundaries, state, and scaffolding retain value.
The Candidate Changed. The Baseline Moves When Trust Moves.
Why agent regression testing has to compare the exact changed candidate against a trusted baseline on the behaviors the workload cannot afford to lose.
The Agent Configuration Became a Release Artifact
Why a stable agent name is not enough to identify the exact configuration that evaluation, debugging, reproduction, and rollback actually apply to.
Capability Needed a Containment Contract
Why capable agents need enforceable limits on execution, network reach, credentials, identities, tools, approvals, persistence, and revocation.
The Diagnostic Build Was Not the Release
A May 2026 Liminal Tarot paywall failure showed why a build instrumented to expose a runtime defect was not yet the artifact that should cross the release boundary.
Shipping Was Not Traction Evidence
Why shipping and launch activity prove output but do not, without separate evidence, prove demand, traction, or distribution fit.