Serialized Publication

OSS Radar

Open source ecosystem analysis. Serialised issues.

13
Issues
#13
Latest
Latest issue
Issue #13
OSS Radar #13: Hob Wrote Correct Files with Network Access Blocked

Hob's local engine wrote the correct JSON and CSV report with networking blocked. I tested its recommended model, setup cost, waiting time and M1 selection rules.

oss-radarlocal-aimacostesting
Archive
Issue #12
OSS Radar #12: Claude Code’s Status Labels Make Their Own Model Calls

Claude Code's tiny agent-status labels can trigger separate model requests. I checked the binary and timer hook, including a case where the workaround silently misses.

oss-radaraiclaude-codetesting
Issue #11
OSS Radar #11: Qwenputer Ran My Program Inside Qwen’s Layers

I compiled Qwenputer's standalone CPU, changed a program's inputs, and disabled the reference interpreter during generation. The model still computed the right result.

oss-radaraitransformerstesting
Issue #10
OSS Radar #10: PaperPilot Left the Secret in My Redacted PDF

PaperPilot's redaction painted a black rectangle and returned success. Its own text-extraction command recovered the secret. I tested the CLI, MCP and a removal control.

oss-radarpdfprivacymcptesting
Issue #09
OSS Radar #09: tsrs Checks VS Code Faster with 56% Less RAM

tsrs checked VS Code in 1.72s using 2.52GB peak RAM, versus TypeScript 7.1 beta's 4.67s and 5.71GB. I also reproduced a fast-mode compatibility difference.

oss-radartypescriptrustcompilersperformance
Issue #08
OSS Radar #08: Bun AOT — Don't Rewrite TypeScript in Rust Yet

Bun AOT made my TypeScript compiler 3.3× faster with 64% less peak memory. A server executable shrank from 121.5 MB to 30.3 MB with AOT and --trim.

oss-radarbuntypescriptrustperformance
Issue #07
OSS Radar #07: AI Citations — Do Agents Preserve the Evidence Behind an Answer?

OpenScholar, STORM, TruLens, and real Codex/Claude runs show where AI citations improve, where they fail, and what makes engineering evidence useful.

oss-radarai-citationsopen-sourceai-agentssource-attributionanalytics
Issue #06
OSS Radar #06: interp-engine, Neuronpedia's New Interpretability Engine, Tested on a Mac

interp-engine, Neuronpedia's new interpretability engine, names 34 hook points once and compares 27 across engines. I audited the code and ran it on a Mac.

oss-radarinterpretabilityneuronpediavllmtransformerlenssteering
Issue #05
OSS Radar #05: Bun 1.4 Made the Runtime the Dependency

Bun 1.4 moves fifteen package jobs into one Rust-based binary. The package graph shrinks, but compatibility and maintenance move into the runtime.

oss-radarbunjavascriptrustruntime
Issue #04
OSS Radar #04: The Agent Multiplexer Is Becoming a Runtime

Herdr gives coding agents durable terminal sessions, agent state, and a control API. Screen-derived semantics remain its limit.

oss-radaragentsterminalherdrmultiplexer
Issue #03
OSS Radar #03: Buzz Is Slack on AIroids

Buzz gives agents signed accounts and shared local compute. Delegation reliability and private code access remain unfinished.

oss-radaropen-sourceagentscollaborationbuzznostr
Issue #02
OSS Radar #02: The Toolchain Is the Moat

How the Astral acquisition reveals the real war being fought in AI coding — and why everyone is talking about the wrong thing.

oss-radaropen-sourcepythonai-toolingrustastraluv
Issue #01
OSS Radar #01: Vercel's Winter 2026 Cohort

A new series: I find a source, go through it, distill what's worth your time, and tell you what I actually think. First up — Vercel's Open Source Program Winter 2026 cohort.

oss-radaropen-sourcevercelresearchdistillation

Dispatch

GK-001

A note when there is signal.

Engineering records, failed experiments, and the life around the work.