ai

plane-mcp-server http mode always builds an OAuth app and crashes without env vars. A 16-line wrapper unlocks header auth with live token checks.

TL;DR: HTTP mode crashed at startup because it always built an OAuth app without env vars. A 16-line wrapper instead calls get_header_mcp() to run a stateless server on loopback 8211 that validates Authorization headers against Plane. This lets local AI agents securely use all 30 tools with no internet exposure or full OAuth setup.

An article's source does not determine its type, the brief's content does. Detect the mode from content, keep one canonical guide, never duplicate enforcement.

TL;DR: The author nearly split their writing guide into two prompt files, one technical and one reflective, before realizing an article's type comes from the brief's content, not its source. A commit can spark a reflection; a dossier can spark a troubleshooting piece. One adaptive guide with two detected modes beats two files that drift.

A local skill and its upstream both said 2.2.1, but the contents had long diverged. How to rebase a skill without inheriting the drift.

TL;DR: A matching version number hid a rewritten SKILL.md with inverted security doctrine. Rebased the file byte-identical to upstream, moving local rules into references docs and a wrapper script. Lesson: never trust frontmatter versions, byte-compare against upstream, and layer customizations on top.

The transcript must carry the hijack sentence verbatim as data. That is exactly how you prove a media file never acted as an instruction.

TL;DR: The author deliberately embedded a prompt-injection sentence into an audio fixture to test a media pipeline's defenses. Counterintuitively, the test requires the hostile transcript to stay verbatim and labeled untrusted, since injection is about text switching from data to instructions. The real defense is a contract proving content never becomes commands, not filtering text or trusting the model's intelligence.

A 70/0/0 acceptance run made me think my agent pipeline was deterministic. It was the runner that deserved the credit; temperature 0 was no guarantee.

TL;DR: A flawless 70/0/0 acceptance run looked like proof of a deterministic agent pipeline, but it really validated the harness, not the model. Temperature zero doesn't guarantee reproducibility either; batch invariance and LLM-as-judge studies show outputs and verdicts still flip. Trustworthy results come from test isolation plus byte-level diffs against golden files, never from an agent grading its own work.

A video file can be perfectly valid and still carry no audio. A nonstate skip keeps the row visible and holds the pipeline until someone decides.

TL;DR: A supposedly broken MP4 was actually valid, just with zero audio streams. The pipeline now runs ffprobe first, stores an audio_track flag, and emits a visible skip that never marks work done. Frames follow a fixed fps cadence with capped extraction, while audio is downsampled to 16kHz mono WAV for transcription.

My transcription pipeline died silently mid-window. A hang turns out to be an exception that must be caught, and every failure has an owner.

TL;DR: After a hung transcription exposed an uncaught TimeoutExpired exception, the author split pipeline failures into infrastructure issues, which silently retry, and content issues, which log a failure event. Broken files follow a retry ladder capped at three attempts, then a visible skip that still blocks completion. Correct exit-code mapping keeps retries from amplifying hopeless failures.

I thought --max-turns was keeping my delegation costs down. It only counts rounds; the flag that actually holds dollars is --max-budget-usd.

TL;DR: A test revealed Claude Code's --max-turns flag doesn't limit spending; a single-turn task cost $0.045 without ever touching its cap of two. By contrast, --max-budget-usd actually halted a run mid-turn once spending passed $0.01. The takeaway: use budget caps to control costs, keeping turn limits only for stopping endless loops.

Changing a filter rule doesn't mean re-running whisper. The filter is a projection of stored stdout, stamped with a sha256 of the rule.

TL;DR: When the profanity filter changed, the author's first instinct was re-running Whisper on every transcript, even though the audio never changed. The fix: treat filtering as a cheap projection of already-cached Whisper output, not a transcription step. Stamp artifacts with a phrase-list hash, so rule changes mean instant re-projection with zero model calls.