← All creators

ahilles107

User

Gravity — a desktop workspace for AI coding agents

5 indexed · 0 Featured · 16 stars · avg score 76
Prolific

Categories

Indexed Skills (5)

AI & Automation Listed

bus-live-test

Test gravityd's cross-session bot communication against a live daemon, in two modes — synthetic (double runtime, drive the MCP bus directly as the bots, free and deterministic) and real (pty runtime spawning billed Claude Code sessions, observing genuine bot behavior). Use when changing send_message/complete_task, the bus guardrails (reply caps, fan-out, hop limits, terminal kinds, deadlines/expiry), delivery, the system prompt's bus sections, or the artifacts flow — or when someone asks to demo, smoke-test, or observe bot-to-bot traffic on a dev daemon.

16 Updated 1 weeks ago
ahilles107
AI & Automation Listed

marketing-screenshot

Capture the real Gravity desktop app for the marketing site using a disposable daemon and staged example data. Real Claude Code sessions are billed and require explicit consent.

16 Updated 1 weeks ago
ahilles107
AI & Automation Listed

release

Cut a new Gravity release — pick the version, land the version-bump PR, tag it, watch the release workflow, and verify the DMG, the gravityd tarball and any configured updater manifest were published. Use when someone asks to release, ship, cut, publish or tag a new version, bump the version, or when a `v*` release run needs monitoring or diagnosing.

16 Updated 1 weeks ago
ahilles107
Testing & QA Listed

terminal-perf-test

Verify terminal rendering performance and correctness (WebGL renderer, bot-switch terminal cache, forced-repaint nudge, replay bottom-pin) by running the real desktop app in Chrome against a local gravityd with the double runtime, seeding heavy ANSI output, and stress-testing switching/scrolling/streaming. Use when changing TerminalPane, terminalCache, the daemon's attach/replay/resize path, or investigating terminal CPU, lag, black screens, or rendering artifacts.

16 Updated 1 weeks ago
ahilles107
AI & Automation Listed

visual-regression

Work with the Ladle visual-regression check — read a failing `visual` CI run, decide whether a pixel diff is a regression or an intended redesign, and adopt CI-produced baselines into the PR. Use when the `visual` workflow fails, when a PR needs updated component snapshots, when adding or changing a `*.stories.tsx` file, or when someone asks to update/refresh/accept UI snapshots.

16 Updated 1 weeks ago
ahilles107

Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.