Attach review scores to recorded Whalesong traces and build comparison pairs — mark a session good/bad/regressed, then diff two runs deterministically. Use when evaluating agent changes, reviewing a run after the fact, or building a small eval set.
中文简介正在准备中。当前显示原文说明。
查看安装方法安装前可以查看交给 AI 代理的指令内容。
Two jobs: leave durable review marks on traces, and diff runs.
whalesong_score writes a Langfuse-style score row onto a trace. Suggested
names stay consistent so later queries aggregate cleanly:
review — categorical: accepted / needs-work / failedrating — numeric 1–5regression — boolean: true means this run was worse than its baselineAlways pass comment with the one-line reason; a bare score is a label
nobody can act on later. Scores are the only write this plugin performs —
confirm the trace id before writing.
whalesong_compare takes two trace ids (or file paths) and returns a
deterministic diff: overall similarity, per-channel and temporal similarity,
deltas in duration / tokens / errors / p95 latency / finding counts, and the
timeline points where the two runs diverged most.
Good pairs:
regression pair,Similarity near 1.0 means the runs' shape matched; it says nothing about output quality — inspect the differing span names the diff reports.
whalesong_compare → report the deltas that matter.whalesong_score both runs so the verdict is durable.Keep the write side boring: one score, one comment, correct trace id.
还没有评价。欢迎分享你的使用感受。
无需登录即可发表。欢迎分享使用感受、提问，或说明在哪些环境下无法运行。
还没有评论。来写第一条吧。
Audit a page or UI against WCAG 2.2 AA. Runs a static HTML scanner and a contrast calculator, then walks the manual keyboard, zoom, screen-reader-name and form checks the tools cannot do, and reports findings with the criterion, evidence, fix and how to verify. Use for accessibility reviews, WCAG checks, contrast questions or "is this accessible".
中文简介正在准备中。当前显示原文说明。
Break a large, parallelizable goal into bounded work units, coordinate existing agent/worktree machinery, integrate, and verify. Explicit-only.
中文简介正在准备中。当前显示原文说明。
Generate a small set of independent candidate solutions in worktrees, judge them against one explicit rubric, and apply the winner only after PASS verification.
中文简介正在准备中。当前显示原文说明。
Read and act on selected Chrome tabs and frames — their real browser, their logged-in sessions. Use for questions about the current page, filling a form the user is on, following a link they mentioned, or checking what a site actually renders. Not for throwaway automation in a clean profile.
中文简介正在准备中。当前显示原文说明。
Deploy or change a Cloudflare Worker with cf. Review config, preflight, build and dry-run; show the target and change, obtain explicit approval, deploy and verify. Covers versions, secrets, rollback and named modes.
中文简介正在准备中。当前显示原文说明。
Look up current Cloudflare documentation (Workers, Pages, KV, R2, D1, Queues, DNS, WAF, platform APIs) through Cloudflare's official remote MCP server. Use when answering Cloudflare questions or writing Cloudflare config.
中文简介正在准备中。当前显示原文说明。