Bulletins

22 August 2026

From past chats

Put a build fingerprint inside each Go binary

For the small Go services and utilities that recur in your work, add a version subcommand or diagnostic endpoint backed by runtime/debug.ReadBuildInfo. Include the Go version, main-module version, VCS revision, modified-tree flag, and the few build settings that matter. This would make “which executable actually ran?” answerable from the artefact itself, without inventing a parallel versioning system. A useful follow-up question is whether the same fingerprint should also appear once at startup in structured logs.

Give agent-built release artefacts verifiable provenance

You have already considered review cost for agent-written code; the next boundary is the compiled output. Try artifact attestations on one low-risk CLI release so a reviewer can verify the repository, commit, workflow, and digest that produced the binary. Keep the experiment deliberately small: generate an attestation in CI, verify it after download, and record whether this answers a real trust question or merely adds ceremony. If it proves useful, the follow-up is deciding which artefacts warrant this treatment rather than enabling it everywhere.

From the lmika.org front page

Turn the memorable game theme into a vertical-remix prototype

If the Godot music is surviving overnight, preserve that hook and test how well it carries gameplay variation. Export two or three perfectly aligned stems—perhaps rhythm, harmony, and melody—then use AudioStreamSynchronized to fade layers as intensity changes. Compare that with AudioStreamInteractive, which is better suited to moving between distinct musical sections. A focused question for the prototype is whether changing orchestration alone gives enough variety, or whether the game genuinely needs section-to-section transitions.

Make an Australian wildlife-risk note from data, not folklore

The gap between overseas fear and ordinary Australian experience could make a sharp little research note. Start with the latest AIHW figures, separate hospitalisations from deaths, and compare venomous animals with the more mundane animal contacts that actually send people to hospital. Pair the numbers with a compact “rare does not mean ignore it” box containing authoritative snake and spider first aid. The interesting follow-up is which statistic best corrects the myth without trivialising the cases that do occur.