{
  "markdown": "# my quant doesn’t speak english\n\nThe public editorial and advertising front door for the Liquidity Lab family:\nSeiche, LiquiLens, and LiquiLens—Undertow.\n\nThe site consumes each product's structured editorial feed at build time. Its\n`SOURCE_PUBLISHED` website mode gives every record the source marks published a\nMyQuant reading page, while preserving the specialist's canonical URL, exact\nclaim, evidence clocks, fingerprint, and limitation. The mobile app feed is\nindependently `SUSPENDED` and contains zero stories.\n\nEvery specialist page must also pass the analysis-quality gate. It answers what\nthe source says, what changed (or why a comparison is unavailable), why the\nmechanism matters, what to inspect next, and where the evidence stops. The JSON\nFeed carries the same change/next-check fields and its `PASSED` receipt. A build\nfails instead of publishing thin labels, missing source identity, missing clocks,\nor known invalid comparison grammar.\n\n`data/fleet-registry.json` maps the complete registered family at an immutable\nLiquiLens `family.toml` revision: seven core repositories, six research modules,\nand ten adjacent surfaces. Coverage does not imply ingestion. Only LiquiLens,\nSeiche, and Undertow market-liquidity currently publish compatible direct\nreadings; upstream, research, private, paused, delivery, and separate-domain\nprojects remain visible at `/coverage` with their exact reason.\n\nThe public site has two explicit lanes:\n\n- `Interpreted`: plain-English readings of Seiche, LiquiLens, and Undertow work;\n- `MyQuant Analysis`: independently sourced house reporting on important news.\n\nThe Interpreted lane includes reviewed longforms and LiquiLens historical case\nfiles. Case files display each lens's `HIT`, `MISS`, or `VOID` separately and\nstate whether the replay was reconstructed later. They are not silently\npromoted into real-time calls or validated-backtest evidence.\n\nOriginal house reporting lives in `content/` and must declare its sources,\ncontribution, limitations, and publication status.\n\n```bash\nnpm test\nMYQUANT_USE_REVIEWED_CACHE=1 npm run build\nnpm run verify\nnpm run dev\n```\n\nThe build has no runtime dependencies. If a source is temporarily unavailable,\nit uses the last successful `data/cache.json` slice for that source and prints the\ncoverage state in the generated page. The LiquiLens desk channel is additionally\npinned to `liquilens.desk-bit-feed.v1`; a missing or changed root schema is treated\nas an upstream contract failure, so the build keeps the last accepted channel and\nmarks LiquiLens as cached/degraded instead of normalizing an unknown shape.\n\n## Public API and MCP\n\nThe product exposes read-only discovery without accounts or user state:\n\n- `GET /api/v1/capabilities` describes the editorial, feed, and evidence surfaces;\n- `GET /api/v1/health` returns a non-sensitive compatibility heartbeat;\n- `POST /mcp` serves five stateless, read-only tools: discovery, health,\n  bounded latest stories, exact stable-ID retrieval, and evidence-aware search;\n- `GET /openapi.json` is the curated REST contract;\n- `GET /.well-known/ai-catalog.json` and `GET /.well-known/mcp.json` expose\n  public AI/MCP discovery, with the Registry manifest at `GET /server.json`.\n\nThe product version (`0.1.0`), REST contract (`myquant.editorial/1.1`), and\nMCP server (`2.1.1` available) stay distinct. Content tools read the canonical\npublic `/feed.json`, validate its full publication/evidence contract, retain\nsource and release fields, and cap each list/search response at 20 records.\nUse `latest_stories` for a chronological overview, `search_stories` to find\nrecords by text, and `get_story` with a stable ID returned by either list tool.\nSearch normalizes Unicode and whitespace, matches every space-separated term\nas a case-insensitive substring, and returns matches newest first. It does not\nperform semantic search or interpret quotation marks and Boolean operators.\nThe tools return the archive's evidence clocks and limitations; a successful\nhealth check does not establish that its market evidence is current.\nPublic health also reports Vercel's exact deployed Git SHA when available,\nallowing scheduled smoke and Registry publication to bind the live endpoint to\na reviewed commit.\n\nRegistry verification builds with `MYQUANT_USE_REVIEWED_CACHE=1`; this uses\nthe committed evidence receipt instead of pulling a different editorial wire\nafter the release SHA has already been selected.\n\nThe MCP endpoint supports the current per-request protocol metadata as well as\nthe listed legacy protocol handshakes. Legacy `Mcp-Session-Id` headers are\naccepted for compatibility but never create server-side sessions. Cross-origin\nMCP requests are limited to the configured MyQuant site origin; the REST\nmetadata remains public and cacheable.\n\nThe deployed Streamable HTTP endpoint is\n[`https://myquantdoesntspeakenglish.com/mcp`](https://myquantdoesntspeakenglish.com/mcp).\nNo API key is required for these public read-only tools. Connect through the\n[Glama remote connector](https://glama.ai/mcp/connectors/io.github.beepboop2025/myquant-editorial)\nor a compatible MCP client. The separate\n[Glama repository listing](https://glama.ai/mcp/servers/beepboop2025/myquantdoesntspeakenglish)\ntracks source licensing and optional Glama-hosted releases; a repository scan\nis distinct from inspection of the deployed remote connector.\n\n## Software license and content rights\n\nFirst-party software is available under the [MIT license](LICENSE). The license\ndoes not cover editorial articles, collected data, third-party content, media,\nor branding. Fonts retain their existing SIL Open Font License notices.\n[RIGHTS.md](RIGHTS.md) describes these boundaries. Installing the software or\nreading the MCP does not grant additional rights to its underlying sources.\n\n## Reviewed shadow copy\n\nModel output never enters the network build or scheduled sync path. An operator\nmay import one immutable, already-validated candidate with:\n\n```bash\nnpm run app-copy:import -- \\\n  --packet /absolute/path/evidence-packet.json \\\n  --candidate /absolute/path/draft.json \\\n  --validator /absolute/path/validation.json \\\n  --review /absolute/path/public-copy-review.json\n```\n\nThe importer checks exact file hashes, source ID and fingerprint, evidence\npacket and validator bindings, support pointers, shadow-only authority, immutable\nteacher/model revisions, and two-reviewer adjudication. A source revision makes\nthe copy ineligible immediately: the website uses its deterministic\n`SOURCE_GROUNDED` explanation and the app edition omits the record. See\n`docs/APP-COPY-IMPORT.md` for the review receipt contract.\n\nThe scheduled `sync evidence wire` workflow checks seven upstream publication\nchannels at\n02:17, 08:17, 14:17, and 20:17 UTC. A changed evidence fingerprint produces one\nsmall content commit, which becomes the deployment trigger; an unchanged network\nproduces no commit and no redeploy. Frequency is a check cadence, not a quota.\n\n## Editorial rule\n\nAutomation may collect and normalize published records. It may not manufacture a\nfact, silently infer calm from a missing feed, or publish a house investigation\nwithout the source and evidence fields described in `content/README.md`.\n\nWebsite archive mode is an operator distribution instruction, not a claim that\nevery source record received legal or regulatory clearance. Corrections,\nretractions, canonical links, content boundaries, and the emergency stop remain\nactive. The homepage uses a first-party silent cinema cut derived from the\napproved app launch creative and documented in `assets/media/`. Third-party\nfield recordings are explicitly excluded from the published build.\n",
  "bytes": 7665,
  "sha": "31dd9e6a3c4fc7e24c8644c112560925b5ae2ad84b0d1065f3e7007fe3f8bd8a",
  "repo_slug": "beepboop2025/myquantdoesntspeakenglish",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/mcp_io_github_beepboop2025_myquant_editorial_7355dd83/readme"
}