Recorded case

Keep VECTORBASE, evaluate retrieval

Use a private source-linked index to find evidence; do not treat search results as authority.

Reviewed 15 October 2026 · scope: implemented

Recorded method and findings

The retained 14 October index build reported 114 documents. The retrieval receipt also reported a failed top-eight target check, even though a targeted query returned the handoff. That is a reason to test retrieval quality, not to discard the source collection or declare search solved.

Recorded method and findings

Our preferred unit is a short question-specific knowledge card linked to the original path, source hash and observation date. Raw chat archives can contain repetition, secrets and abandoned advice. They should be curated into evidence and decisions before indexing or publication.

Practical checklist

Keep one canonical retrieval route and avoid competing indexes of the same meaning. Evaluate known questions and negative cases before relying on search for decisions. Reopen the original source and check its version before acting.

Sources & scope

Internal retrieval index build receipt 2026-10-14
Internal retrieval evaluation receipt 2026-10-14
Internal retrieval source and freshness contract 2026-10-15

Keep exploring

AI orchestration: from a prompt to an accepted resultA practical design for delegation, evidence and recovery.Shared memory without shared mistakesKeep facts, hypotheses and permissions distinct across agent sessions.Engineering that can show its workTVIJO connects operators, bounded agent work and knowledge that can be checked and reused.AGENTORCH: the orchestrator has a concrete jobAGENTORCH names the orchestration role: plan, assign, observe, reconcile and accept.The task packet is the interfaceA good task describes a verifiable outcome rather than an invitation to improvise without limits.AGENTJUDGE: acceptance follows evidenceA reviewer checks the task, the artifact version and each acceptance criterion.Measure cost per accepted resultSmall task benchmarks illustrate why task class and acceptance matter more than a cheap model label.Budgets and outages belong in the state machineA reliable operator view distinguishes working, waiting, failed, blocked and accounting uncertainty.SPIDER: evidence collection and useful progressThe first implementation reads an explicit project-evidence list and produces a dated operator digest without model calls.DSHREMBOT and tmux: one operator viewAuthenticated project commands, budget visibility and explicit schedule controls connect the operator to existing execution lanes.The AI worker–Hugging Face incident: learn from controlsThe incident reports support specific engineering lessons, not a controlled ranking of our system.What the supplied AI worker screenshot actually provesThe screenshot records a safeguard-triggered switch in one session; it does not explain the exact trigger. The observation is the supplied screenshot dated 15 October 2026.Training our own model: a gated research directionA proprietary model is an option to test after retrieval, prompts and routing have measurable limits.A bilingual knowledge product with visible limitsPublish useful explanations first, then expand languages when the evidence and workflow are ready.Free learning, custom projects and future challengesTurn evidence-backed engineering into clear offers and a useful developer community.Trust the sender, not a public checksumBind the complete message to an explicitly trusted authority.