Codex

Output defaults to readable text with JSON, JSONL, raw and YAML options, and --transform extracts fields with GJSON syntax. Setting OPENAI_UNTRUSTED_STDIN=1 blocks reading files from piped input.

Install by Homebrew, release binary or Go 1.26.

An Ask HN thread asked how engineers cope with shipping LLM-written code they haven't fully read. A Hasura engineer says his own system became fuzzy once he moved from IDE autocomplete to Claude and Codex CLIs, and argues the answer is formally verifiable code.

The loudest counter: you already ship code you haven't read. Principal engineers always have.

DeepSeek published Search and Fetch APIs at $0.75 per 1,000 successful requests, billed only on HTTP 200 responses. Calls routed through Claude Code, Codex or DeepSeek's own harness are free until 00:00 Beijing time on November 1, 2026.

The first reply was a price check: Kagi's API runs $12 per 1,000 searches. Then a hobbyist who built his own agent tools said the big labs' bundled search costs 15-20x more than his.

The experiment paired an agent with benchmarking, profiling, research, and correctness verification. That made optimization a constrained search problem.

HN readers reported that agents work best where a system supplies a hard answer. They fail more readily when success depends on credentials or a human-operated dashboard.

The preview supports Ubuntu 24.04 and 26.04 LTS, Debian 13, Fedora 43 and 44, on x64 and ARM64 through deb and rpm packages.

Early users immediately tested the boundary: Wayland input methods, downloads from another OS, and whether desktop projects can see existing CLI work.

The complaints ran from petty (re-reading its whole config file every turn, wrecking prompt caching) to serious: approving one throwaway Python command with "Always" silently green-lights every future Python call, including one that reads your SSH keys.

Simon Willison defended one pettier complaint, baking the date into every prompt, as standard practice. OpenCode's own developer said the harshest bug, a context-pruning behavior, was already disabled by default months before the post went up.

The announcement is thin — Ultra reportedly uses subagents to go beyond a single agent. The thread fills with access logistics and a guess that Pro-tier already runs ~12 parallel agents synthesized into one answer.

The reversal is a corporate weather report: praise for the biggest token users flipped to weekly emails begging for restraint — arriving just as OpenAI reportedly found a way to halve inference cost.

Across 390,195 logged responses, GPT-5.5 Codex terminates reasoning at exactly 516 tokens far more than any sibling: 82% of exact-516 events are GPT-5.5, though it's only 19% of traffic. The stops cluster at 516, 1034, and 1552, each about 518 apart.

One HN user reproduced it live: 4 of 10 identical prompts short-circuited at 516 tokens, and every one returned the wrong answer.

The suspected cause: a throughput optimization batching reasoning in multiples of 512.

Plans run from $16.20 to $144 a month, and GLM Coding subscribers get a quota discount inside ZCode: peak hours (14:00-18:00) drop from 3x to 2x and off-peak from 1x to 0.67x, roughly a 1.5x effective allowance. The base allowance itself is never stated in tokens.

Developers ask why a lab that open-weights its models keeps the harness closed, and a Chinese-owned agent with full system permissions drew explicit trust warnings.

Adrafinil, a Show HN named after a modafinil derivative, keeps a lid-closed Mac awake only while AI agents are working, then sleeps. Useful — until the thread points at the wheel it reinvents.

macOS ships 'caffeinate -dimsu.' Amphetamine has done conditional triggers for years. One commenter compressed the whole thing to a three-line shell loop; another asked why a sleep toggle needs 15,000 lines of code; a third noted the coding agent itself just shipped the feature.

Another Anthropic outage hit mid-session: 500s, then 529 Overloaded, then a status page one user compared to Christmas ornaments. A reader scraped the page's own incident data and recomputed the real figure: 97.68% uptime over 90 days, roughly 50 hours of degraded service.

The sharper thread isn't the downtime. It's developers asking how a shop that says it runs on agent loops fixes its own infra when the agent is the thing that's down.

TikZ Editor lets you draw figures visually and outputs TikZ/LaTeX code, open source and on the web. The most-praised use isn't drawing from scratch — it's editing old TikZ without turning the source into generated soup.

Maker DominikPeters says it took 700M tokens since February via a ChatGPT subscription. A LaTeX user in the thread flagged the catch: every node lands at absolute coordinates, which is rarely how good TikZ is written.

The SQLite log kept only ~500K rows, yet its autoincrement counter had passed 5.5 billion: a frantic insert-and-prune churn, 70.7% of it raw TRACE noise. The cause was a one-line default capturing every diagnostic at Level::TRACE. A maintainer who'd hit an identical bug years before pushed back on the slop jokes, then someone found Claude Code doing the same thing.