Building Glasstrace

Ten articles documenting the work on Glasstrace from April through August 2026: the reasoning behind the major decisions, the architecture trade-offs, and the validation program. Published .

  1. Part 1 Choosing Development Time Over Production: The Founding Constraints Why Glasstrace runs only at development time: the founding constraints and what they bought.
  2. Part 2 Validating the Zero-Configuration Claim: An Installation Audit Across Six Codebases An installation audit of the zero-configuration claim across six real codebases.
  3. Part 3 Synthetic Evidence in the Validation Pipeline: The June Capture-Gap Finding How the validation pipeline discovered its most convincing evidence was synthetic, and what changed.
  4. Part 4 Diagnosing a Ten-Millisecond Span-Export Race in Next.js Diagnosing a ten-millisecond span-export race in Next.js Server Actions.
  5. Part 5 Preregistration as Policy: The Eighty-Pair Study Preregistration as standing policy: the eighty-pair study and its findings.
  6. Part 6 The Scarcity of Admissible Bugs in Agent-Debugging Research Why admissible bugs are the scarce resource in agent-debugging research.
  7. Part 7 Validation Infrastructure Before Features: Building the Study-Execution System Building the study-execution system before adding features.
  8. Part 8 An Unresolved Instrument Effect: The Observability Note An unresolved instrument effect: how the observability note changed agent engagement.
  9. Part 9 The Design and Execution of Study E The design and execution of Study E.
  10. Part 10 Post-Hoc Analysis Under a Frozen Protocol: Mining Study E Mining Study E under a frozen protocol: post-hoc analysis with guardrails.

Every factual claim in the series is sourced; the evidence manifest lists the cited commits and the SHA-256 digests of the supporting artifacts.