OpenCode adds built-in which-key plugin

// 1h agoPRODUCT UPDATE

OpenCode adds built-in which-key plugin

The upcoming OpenCode release adds a built-in which-key plugin that shows the currently active keybindings at any time, making the terminal UI easier to discover and use. The post is a repost of a short teaser, but the core signal is clear: OpenCode is continuing to polish its TUI ergonomics for power users who rely on keyboard-driven workflows.

// ANALYSIS

Hot take: this is a small feature with outsized UX value, especially for a terminal-first tool where hidden shortcuts can become a real adoption barrier.

–Improves discoverability without changing the core workflow.
–Fits OpenCode’s audience: users who want fast keyboard navigation and low-friction command access.
–Suggests the project is iterating on polish, not just raw capability.
–Likely more useful for new users than veterans, but still handy when keybind sets change across modes.

// TAGS

opencodewhich-keyterminal-uikeyboard-shortcutsdevtoolopen-source

DISCOVERED

1h ago

2026-05-09

PUBLISHED

2h ago

2026-05-09

RELEVANCE

7/ 10

AUTHOR

thdxr

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

NEWS2h ago

Anthropic’s SpaceX deal lifts Claude limits

Theo’s video covers Anthropic’s May 6, 2026 announcement of a compute partnership with SpaceX. The deal expands Claude capacity and raises Claude Code and Claude Opus limits.

BENCHMARK2h ago

ClickUp agents top ChatGPT, Claude evaluations

ClickUp’s benchmark report says its Certified Agents scored 96/100 and outperformed ChatGPT with connectors, Copilot, Notion agents, and Monday agents on execution-ready project planning. The claim is really about workflow orchestration and context inside the work system, not raw model intelligence.

OPEN SOURCE2h ago

iFixAi launches an open-source diagnostic for AI misalignment with 32 provider-agnostic tests and replayable scorecards.

iFixAi is a free, open-source diagnostic for AI misalignment that runs 32 inspections across five risk categories: fabrication, manipulation, deception, unpredictability, and opacity. It works across major providers including OpenAI, Anthropic, Azure OpenAI, Gemini, Bedrock, Hugging Face, OpenRouter, HTTP, and LangChain, and it produces a letter grade in under five minutes along with a content-addressed manifest for deterministic replay. The project is positioned as a repeatable CI signal and comparison tool rather than a safety certification, with outputs designed to be tracked over time and audited later.