[{"_1":2,"_3":-5,"_4":-5},"loaderData",{"_5":6,"_7":8,"_9":10},"actionData","errors","root",{"_117":118},"routes/modules.$creator.$moduleId",{"_14":15,"_16":17,"_18":19,"_20":21},"routes/modules.$creator.$moduleId.endpoints",{"_11":12,"_13":-5},"error","Module not found","tools","module",{"_20":22,"_23":21,"_24":25,"_26":27,"_28":29,"_30":31,"_32":33,"_34":25,"_35":36,"_37":-5,"_38":-5,"_39":-5,"_40":41,"_42":43,"_44":45,"_46":47,"_48":49,"_50":51,"_52":53,"_54":-5,"_55":47,"_56":57,"_58":59,"_60":-5,"_61":-5,"_62":-5,"_63":-5,"_64":65,"_66":67,"_68":69,"_70":71,"_72":73,"_74":-5,"_75":-5,"_76":-5,"_77":47,"_78":-5,"_79":80,"_81":82,"_83":84,"_85":86,"_87":88,"_89":90},"apiUrl","https://odel.app/api","creator","attenlabs","moduleId","hotato","@attenlabs/hotato","displayName","tagline","Open-source, self-hosted conversation QA for voice agents. MIT.","iconKey","https://avatars.githubusercontent.com/u/67333393?v=4","category","local","status","published","visibility","public","metaDescription","deploymentType","external","workerName","bindingName","mcpConfig","pricingModel","free","creatorUserId","system:external-import","createdAt",1783571141000,"updatedAt",1788329317000,"sourceRegistry","official-mcp-registry","sourceUrl","https://registry.modelcontextprotocol.io/v0/servers?search=io.github.attenlabs%2Fhotato","sourceVerified",true,"sourceMissingSince","importedAt","upstreamStars",1,"upstreamUpdatedAt",1787929082000,"detectedTransport","detectedAuthType","probeStatus","lastProbedAt","githubRepoUrl","https://github.com/attenlabs/hotato","githubRepoFullName","attenlabs/hotato","releaseId","ext-attenlabs-hotato","version","1.9.0","descriptionMarkdown","

\n\n $\"hotato\"$ \n\n

\n $\"PyPI$ \n $\"Downloads$ \n $\"Python$ \n $\"CI$ \n $\"MIT$

\n\n\n# hotato\n\n**Find what broke in your agent calls. Pin it and CI stays red until you fix it.**\n\n```bash\npip install hotato\nhotato check --demo # a bundled failing call: what broke, and when\nhotato check ./call.wav # then your own recording\nhotato vapi health # or your last 100 Vapi calls\n```\n\nZero config. Vapi, Retell, Bland, Synthflow, Millis, or local audio.\nWorks with any recording: mono, dual-channel, or a transcript. Timing math, not a judge. Offline. Free. MIT.\n\n**[hotato.dev](https://hotato.dev)**\n\n

\n\n## What it finds\n\n- **Say-do gaps**: the caller interrupts to cancel (a barge-in), the agent says \"canceled\", the booking tool fires anyway. hotato takes the turn timing from the audio and the tool call from your OTel trace.\n- **Latency spikes**: the pause before a reply going from 800 ms to over 2 s.\n- **Dead air**: that pause reaching 5 s, or the line going quiet.\n- **Talk-over**: the agent starts a fresh utterance over the caller.\n\n## Quickstart\n\n### Vapi\n\n```bash\npip install hotato\nexport VAPI_API_KEY=...\nhotato vapi health --last 7d --output report.html\n```\n\nOpen `report.html`: every critical incident, timestamped, and your Voice Stability Score.\n\n### Retell\n\n```bash\nexport RETELL_API_KEY=...\nhotato retell health --call-id CALL_ID\n```\n\n`--call-id` is required and repeatable: you name the Retell calls to pull.\n`hotato bland health`, `hotato synthflow health`, and `hotato millis health`\nfollow the Vapi shape. Those stacks mix both voices onto one channel, so\nthey measure silence timing, dead air and latency gaps, each finding with\nits measured confidence; barge-in and talk-over need two channels.\n\n### Local audio\n\n```bash\nhotato autopsy ./call.wav\n```\n\nWrites a self-contained HTML report to `hotato-output/`, plus the JSON\n`pin` reads. Open the HTML in your browser.\n\n## From finding a bug to gating on it\n\n`autopsy` turns one recording into timestamped incidents. `scan` reads a\nfolder and tracks the trend. When you are ready, move a finding into CI:\n`hotato pin` turns one incident into a portable failure check, and `hotato\nprove` re-runs every stored check and fails the build rather than pass on\nevidence it cannot re-read. Every verdict carries its own evidence across\nfive dimensions: outcome, policy, conversation, speech, reliability.\n\nA pinned check re-measures its own stored recording, so it holds the build\nred on that bug and catches an edited bundle, a loosened policy, or an\nengine upgrade that scores the same bytes differently. Clearing it takes a\nfresh recording of the same moment against the current agent: `hotato drive\n` places that call on Vapi or Twilio, and\n[`docs/RECAPTURE.md`](https://github.com/attenlabs/hotato/blob/main/docs/RECAPTURE.md) is the walkthrough for every other\nstack.\n\n[Pin a bug →](https://github.com/attenlabs/hotato/blob/main/docs/CI.md)\n\nFor continuous use: run `hotato vapi health` on a schedule, and open\n`hotato console --production-db evidence.db` to watch calls land live.\n\n## Wire it into CI\n\nThe exit code **is** the verdict: `0` pass, `1` fail, `2` refuse (could not tell).\n\n```yaml\n# .github/workflows/voice-qa.yml\non: [pull_request]\njobs:\n hotato:\n runs-on: ubuntu-latest\n steps:\n - uses: actions/checkout@v4\n - uses: attenlabs/hotato@v1.20.0\n with:\n contracts: contracts/\n hotato-version: 1.20.0\n```\n\n`contracts/` holds what `pin` wrote. Full workflow: [`docs/CI.md`](https://github.com/attenlabs/hotato/blob/main/docs/CI.md).\n\n## Point your agent at it\n\nPoint Claude Code, Cursor, or any coding agent at this repo: it reads\n[`AGENTS.md`](https://github.com/attenlabs/hotato/blob/main/AGENTS.md) and runs the loop end to end offline, no key. Over local\nstdio the MCP server adds the scorer plus read/verify/propose tools:\n`uvx --from \"hotato[mcp]\" hotato-mcp` ([`docs/MCP.md`](https://github.com/attenlabs/hotato/blob/main/docs/MCP.md)). Deploying is yours.\n\n## Nothing leaves your machine\n\nhotato runs offline, on the machine that invokes it. The core is stdlib-only\nPython: no account, no key, no network call of its own. Your traces, prompts,\nand audio stay on your disk. Scoring with a language model you host yourself\nis a separate opt-in add-on, outside that core.\n\n## Go deeper\n\nThe whole loop, command by command: [`docs/LIFECYCLE.md`](https://github.com/attenlabs/hotato/blob/main/docs/LIFECYCLE.md).\nFirst touch to a CI gate: [`docs/GETTING-STARTED.md`](https://github.com/attenlabs/hotato/blob/main/docs/GETTING-STARTED.md).\nFeed it what you already have: [`docs/CONNECT.md`](https://github.com/attenlabs/hotato/blob/main/docs/CONNECT.md) ·\n[`docs/TRACE.md`](https://github.com/attenlabs/hotato/blob/main/docs/TRACE.md) · [`docs/SIMULATE.md`](https://github.com/attenlabs/hotato/blob/main/docs/SIMULATE.md).\nWhat every verdict stands on: [`docs/EVIDENCE-CONTRACT.md`](https://github.com/attenlabs/hotato/blob/main/docs/EVIDENCE-CONTRACT.md).\nNext to the hosted alternatives: [`docs/COMPARE.md`](https://github.com/attenlabs/hotato/blob/main/docs/COMPARE.md).\n\nThe deep toolkit -- capture, simulation, load, benchmarking, the fix ladder,\nthe fleet control plane -- lives under `hotato lab` (`hotato lab --help`).\nThe public commands are durable. hotato lab moves faster, and every\ncommand name that worked before 1.17 still runs unchanged.\n\n## Specifications\n\n| Property | Value |\n| :-- | :-- |\n| Footprint | ~10 MiB installed, 0 runtime dependencies (stdlib-only) |\n| Reproducibility | byte-for-byte: the same recording, the same report |\n| Exit codes | `0` pass · `1` fail · `2` refuse |\n| Release integrity | OIDC Trusted Publishing + build-provenance attested |\n| Runtime | offline, off the production data path |\n\n

Verify the measurement yourself

\n\n```bash\nPYTHONPATH=src python3 -m hotato.benchmark \\\n --scenarios corpus/real/scenarios --audio corpus/real/audio\n```\n\nOn 13 recorded AMI Meeting Corpus clips, the median error between measured caller-onset and the human word-alignment label is **20 ms**. Provenance: [`corpus/real/README.md`](https://github.com/attenlabs/hotato/blob/main/corpus/real/README.md) · method: [`METHODOLOGY.md`](https://github.com/attenlabs/hotato/blob/main/METHODOLOGY.md).\n\nTiming is measurable only when the two voices arrive on separate channels; a mono or mixed export is marked **NOT SCORABLE** and refused (`hotato trust --stereo call.wav`). The full four-tier evidence policy (what each verdict stands on, per input) is [`docs/EVIDENCE-CONTRACT.md`](https://github.com/attenlabs/hotato/blob/main/docs/EVIDENCE-CONTRACT.md).\n\n

\n\n## Contribute\n\nIssues and PRs welcome: [`CONTRIBUTING.md`](https://github.com/attenlabs/hotato/blob/main/CONTRIBUTING.md) · [`SECURITY.md`](https://github.com/attenlabs/hotato/blob/main/SECURITY.md) · [`CHANGELOG`](https://github.com/attenlabs/hotato/blob/main/CHANGELOG.md) · [`docs/`](https://github.com/attenlabs/hotato/blob/main/docs/)\n\n## License\n\nMIT ([`LICENSE`](https://github.com/attenlabs/hotato/blob/main/LICENSE))\n\n

_{Know when to pass it on.}

\n\nmcp-name: io.github.attenlabs/hotato\n","changelogMarkdown","videoKey","requiredSecretsJson","publishedAt","priceCents","trialDays",0,"upstream",{"_101":102,"_103":104,"_105":106,"_107":108,"_109":110,"_111":112},"requiredSecrets",[],"tags",[93,94,95,96,97,98,99,100],"assets",[],"stats",{"_91":80,"_92":80},"ratingAvg","reviewCount","call-analysis","forensics","regression-testing","retell","testing","vapi","voice-agents","voice-ai","topics",[93,94,95,96,97,98,99,100],"license","MIT","language","Python","registryName","io.github.attenlabs/hotato","homepage","https://hotato.dev","packages",[113],{"_114":115,"_116":21,"_70":71},"registryType","pypi","identifier","ENV",{"_119":120,"_121":122,"_123":124},"ENVIRONMENT","production","TURNSTILE_SITE_KEY","0x4AAAAAAB8-rEI2gNDf_QhU","SENTRY_DSN","https://45fd1229f78089d5c3ee78be128030e1@o4510583261757440.ingest.de.sentry.io/4510584253448272"]