Quantify periodicity in a recorded agent session — dominant onset period in seconds, spectral entropy, autocorrelation lag. Use to prove a poll loop, measure how metronomic retries are, compare cadence between runs, or window into one suspicious stretch.
The English summary is being prepared. Showing the original description.
See how to installReview the instructions the agent will receive before you install.
whalesong_rhythm bins a trace's onset train into 1,024 samples and runs
the instrument's real DSP: centered variance-normalized autocorrelation and
a Hann-windowed one-sided periodogram. Frequencies here are cycles/second
of the event train — not audio pitch, not the sonogram's semantic rows.
dominantPeriodSeconds — the strongest repetition interval. ~4s means
something fired every four seconds; match it against loop findings.spectralEntropy 0–1 — low = one dominant rhythm (metronomic), high =
arrhythmic activity. Use it to separate "stuck on a timer" from "busy".strongestAutocorrLagSeconds + autocorrAtBestLag — self-similarity at
its best offset; a high value confirms periodic structure the spectrum
might split into harmonics.Default is the whole trace. Narrow with start_ms/end_ms (finding
atMs values are relative to trace start — feed them straight in) or one
category (e.g. tool for just the call cadence, reasoning for model
cadence). A stalled session's gap boundaries make good window edges.
whalesong_analyze or the sonogram flags a repeating pattern.whalesong_rhythm on the flagged window → period + confidence.whalesong_observations on that window, full: true → what the
metronome was actually calling.bash <cmd> every ~8.5s for 40 min",
not "exhibited anomalous periodicity".Cadence is a property of the record — it still doesn't prove intent.
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
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".
The English summary is being prepared. Showing the original description.
Break a large, parallelizable goal into bounded work units, coordinate existing agent/worktree machinery, integrate, and verify. Explicit-only.
The English summary is being prepared. Showing the original description.
Generate a small set of independent candidate solutions in worktrees, judge them against one explicit rubric, and apply the winner only after PASS verification.
The English summary is being prepared. Showing the original description.
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.
The English summary is being prepared. Showing the original description.
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.
The English summary is being prepared. Showing the original description.
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.
The English summary is being prepared. Showing the original description.