Technical Artifact Index
Generated reports and machine-readable outputs for reviewers who want to inspect the evidence behind the public summary.
Source documentation lives in the repository docs folder.
Primary reports
Intervention study
- Baseline v1 summary
- Agent safety intervention study
- Instruction hierarchy intervention
- Instruction hierarchy report
- Action gate intervention
- Action gate report
- Safety classifier intervention study
- Safety classifier intervention report
- Memory context intervention study
- Memory context intervention report
- Goal conflict intervention study
- Goal conflict intervention report
Benchmark profile
Public RAG
- TechQA public RAG summary
- TechQA public benchmark profile
- TechQA public retriever comparison
- WixQA public RAG summary
- WixQA public benchmark profile
- WixQA public retriever comparison
- Cross-public RAG findings
- Public RAG reranking opportunity
- Public RAG reranker evaluation
- RAG grounding intervention study
- RAG grounding intervention report
- Hosted public RAG reranker adapter
- Hosted reranker packet
Model comparison
Safety evaluation
- Safety classifier summary
- Safety threshold sweep
- Safety threshold retuning
- Safety human review simulation
- Safety adjudication notes
- Safety reviewer disagreement slices
- Safety secondary review-band analysis
- Safety secondary review-floor validation
- Safety secondary review operating recommendation
- Safety mitigation impact
- Safety threshold decision memo
Human review
Observability
Incident replay
- Incident replay summary
- Incident replay runs
- Incident release gates
- Incident response plan
- Incident pack schema
- Candidate results schema
- Incident memo INC-2026-0001
- Incident memo INC-2026-0002
- Incident memo INC-2026-0003
- Incident memo INC-2026-0004
- Incident memo INC-2026-0005
- Incident memo INC-2026-0006
- Incident memo INC-2026-0007
- Incident memo INC-2026-0008