The Analyst and The Gossip Girl read the same public record and disagree on purpose. What is this?
Coverage Disclosure

What We Monitor and What We Do Not

A coverage page that only lists what exists is marketing. This page also lists what is deliberately not monitored — the signals this product cannot see, the scrapers that exist but are not wired, and the structural gaps that no amount of configuration would close. Stating the boundary is what makes an intelligence product credible.

31
Scrapers
19
Source Types
11
Not Covered

Scraper Registry

KeyModuleGroupStatusAPI KeyProducesObserved Signals
rssrss-scrapercoreregistered—
RSSPODCAST
8
filingfiling-scrapercoreregistered—
FILING
14
blogblog-scrapercoreregistered—
BLOG
5
socialsocial-scrapercoreregistered—
SOCIAL
8
jobPostingjob-scrapercoreregistered—
JOB_POSTING
9
transcripttranscript-scrapercoreregistered—
TRANSCRIPT
8
newsnews-scrapercoreregistered—
NEWS
36
githubgithub-scraperextendedregistered—
TECH_SIGNAL
8
certTransparencycert-transparency-scrapergovernmentregistered—
TECH_SIGNAL
8
redditFinancialreddit-financial-scraperextendedregistered—
SOCIAL
8
pressReleasepress-release-scraperextendedregistered—
PRESS_RELEASE
6
usptouspto-scrapergovernmentregistered
PATENT
7
courtListenercourtlistener-scrapergovernmentregistered
LITIGATION
5
fdafda-scrapergovernmentregistered—
FDA
1
samsam-scrapergovernmentregistered
CONTRACT
5
waybackwayback-scraperextendedregistered—
WEB_ARCHIVE
6
congresscongress-scrapergovernmentregistered
LEGISLATION
8
academicacademic-scraperextendedregistered—
ACADEMIC
3
lobbyinglobbying-scraperextendedregistered—
LOBBYING
6
supplierEarningsupplier-earning-scrapernewregistered—
TRANSCRIPT
8
execAppearanceexec-appearance-scraperextendedregistered—
CONFERENCE
5
appStoreTrackerappstore-trackertrackerregistered—
TECH_SIGNAL
8
domainTrackerdomain-trackertrackerregistered—
TECH_SIGNAL
8
conferenceAgendaconference-agenda-scrapernewregistered—
CONFERENCE
5
appStoreapp-store-scrapernewregistered—
TECH_SIGNAL
8
conferenceconference-scrapernewregistered—
CONFERENCE
5
webSearchweb-search-scrapernewregistered—
NEWSBLOGJOB_POSTINGFILINGSOCIALPATENTPRESS_RELEASETRANSCRIPT
93
stealthBrowserstealth-browsernewregistered—
NEWS
36
mastodonmastodon-scrapernewregistered—
SOCIAL
8
twittertwitter-scrapercorepresent-unregistered—
SOCIAL
8
jinaReaderjina-readercorepresent-unregistered—
NEWSBLOG
41

Source Types

Source TypeSignal CountProduced By
NEWS36
news-scraperweb-search-scraperstealth-browserrss-scraper
FILING14
filing-scraperweb-search-scraper
TRANSCRIPT8
transcript-scrapersupplier-earning-scraperweb-search-scraper
SOCIAL8
social-scrapertwitter-scrapermastodon-scraperreddit-financial-scraperweb-search-scraper
BLOG5
blog-scraperweb-search-scraper
JOB_POSTING9
job-scraperweb-search-scraper
RSS2
rss-scraper
PATENT7
uspto-scraperweb-search-scraper
LITIGATION5
courtlistener-scraper
FDA1
fda-scraper
CONTRACT5
sam-scraper
TECH_SIGNAL8
github-scrapercert-transparency-scraperapp-store-scraperappstore-trackerdomain-tracker
WEB_ARCHIVE6
wayback-scraper
LEGISLATION8
congress-scraper
ACADEMIC3
academic-scraper
PODCAST6
rss-scraperweb-search-scraper
CONFERENCE5
conference-scraperconference-agenda-scraperexec-appearance-scraper
PRESS_RELEASE6
press-release-scraperweb-search-scraper
LOBBYING6
lobbying-scraper

What We Deliberately Do Not Monitor

11 items the product cannot see — each grounded in a specific behaviour of the code rather than a general statement of limitations.

10-K exhibits

filing-scraper records only primaryDocument and primaryDocUrl from the EDGAR submissions index and ingests that single document as rawContent. Exhibit 99.x press releases, the segment footnote tables, the commitments-and-contingencies note, and the auditor consent pages are separate documents in the same accession and are never fetched. For most issuers the most quotable disclosure in a 10-K is in an exhibit, so the numbers this product treats as primary-tier evidence are structurally incomplete for exactly the filings that matter most.

Blocked by: Walking the EDGAR accession index and ingesting each exhibit document, then attributing exhibits to their parent filing rather than treating them as standalone signals.

XBRL company facts

Nothing in the codebase calls data.sec.gov/api/xbrl/companyfacts, companyconcept, or the frames API — those endpoints appear nowhere outside this comment. Every financial figure reaching a signal arrives as prose inside a document body and is extracted by the analysis agents, which means a restatement or a segment reclassification shows up as new narrative rather than as a revision to the number it replaced. Two sources of financial truth would also be unavoidable if this were wired: the filing text and the tagged facts, which disagree in practice.

Blocked by: A facts ingestion path keyed on CIK, a tag-level time series model, and a reconciliation step that records where tagged facts and narrative extraction diverge.

Insider Form 4 detail

Form 4 is in TRACKED_FILING_TYPES so the scraper sees the filing in the submissions index, but the ingest path stores the primary document as unstructured text. Transaction codes, share counts, prices, and post-transaction holdings are not parsed, and no ownership graph is built. A Form 4 signal in this product is therefore evidence that a filing exists, not evidence of what was bought or sold — which is the only part anyone would want it for.

Blocked by: A structured Form 4 parser producing transaction-level rows, plus a holder-level model to resolve the reporting owner to a person the people corpus already tracks.

13F holdings

13F is tracked at the index level for the same reason as Form 4, and runs into the same wall: the information table in a 13F is a structured exhibit, not narrative in the primary document, and holdings are reported 45 days after quarter end. No manager-to-position table is built, so there is no way to ask which funds added a position or when a holding first appeared. A 13F here is a filing notice, not a position.

Blocked by: Exhibit ingestion, a manager identity resolution step, and quarterly position diffing to turn individual 13Fs into ownership-change signals.

Short interest

There is no short-interest source in the pipeline. reddit-financial-scraper mentions short interest in a comment describing what retail discussion implies about it, which is sentiment about the topic rather than the data. Nothing polls FINRA bi-monthly dissemination or any exchange equivalent, so a crowded-short thesis cannot be evidenced at all — and the only place the subject appears is inference from retail chatter, which is exactly the kind of weak claim this product should refuse to make.

Blocked by: A FINRA short-interest feed with settlement-date semantics, plus historical series storage so a move can be distinguished from a level.

Options flow

No options data source is integrated anywhere — no chain, volume, open interest, or block-trade collection. The implication is that a company about to be targeted by an options-based strategy is indistinguishable from one that is not, at the moment it would matter most. Any inference about positioning from price action in this product would be invented, so none is attempted.

Blocked by: A licensed options feed (OPRA or equivalent) with block-trade and sweep classification, which is a commercial data contract rather than a public scrape.

State and local government records

sam-scraper polls SAM.gov, which is US federal procurement only, and congress-scraper polls Congress.gov. No state or municipal contract register, no state legislative tracker, and no state lobbying database is wired, even though the corpus contains a Storey County commission record and a Nevada state water permit feed that were found by general web search rather than collected by a dedicated scraper. That distinction matters: search-found records arrive unvalidated and stop the moment a site reorganises, which is precisely how the retired committee feed in the source-health table went dark.

Blocked by: Per-state integrations, a normalised entity resolver to match a county or state body to a tracked company, and validation semantics as strict as the federal sources.

Private-company financials

Almost every discovery path gates on a ticker: eight separate steps in signal-discovery.ts filter to companies where ticker !== null before collecting, and the filing path needs a CIK. A private company has neither, so it is excluded from filings, patents-by-ticker lookups, academic, lobbying, supplier earnings, and executive appearances. Private-company intelligence here can only come from news, jobs, blogs, and social — which is real signal about intent and hiring, and is not a substitute for the financials.

Blocked by: Entity resolution for private companies (legal name, aliases, parent/subsidiary relationships), a document-first pipeline that does not require a CIK, and a source class for private disclosure that does not exist.

Non-US issuers

Coverage is structurally US-shaped. Filing collection is EDGAR-only; a foreign private issuer's annual report on Form 20-F is not in TRACKED_FILING_TYPES, and ASML — the one non-US issuer tracked here — reaches this product through news, patents, and conference schedules rather than through a primary filing source. Government sources are all US agencies. A Chinese or Korean issuer is covered only to the depth that English-language web search finds.

Blocked by: National filing-registry integrations ( Companies House, EDINET, national equivalents), a per-jurisdiction document-type mapping, and entity resolution across transliteration and local-language names.

Real-time price data

This is a deliberate design decision, not a gap, and stating it as a gap would misrepresent the product. Price and quote feeds are not collected: no market data vendor is integrated, and the only price-like values in the pipeline are App Store app prices scraped as product metadata. A signal in this product is always anchored to a document with a publication date, and a real-time quote would attach an implied urgency to an hours-old filing that the document does not support. Readers who want price action should treat the publication timestamps as the temporal frame and read the feed accordingly.

Blocked by: Nothing technical — a market data contract if it were ever wanted. The reason to decline is editorial: a document-anchored signal should look like a document, not like a ticker tape.

Paywalled and JS-rendered article bodies

Two sources on the health table illustrate the cost. Financial Times returns 451 on article bodies under its crawler policy, so this product collects headlines and drops bodies. Where a page renders client-side, the fast HTTP tier yields an empty shell and the fallback chain escalates to a self-hosted Jina Reader and then a stealth browser — which works but is slow, rate-limited, and absent from the registry, so it cannot be enabled, disabled, or monitored as a first-class source. Paywalls are not bypassed; those sources are reported as degraded.

Blocked by: Publisher licences for the paywalled sources, and a registry entry plus health coverage for Jina Reader and the stealth browser so escalation cost is visible rather than invisible.

Scrapers That Exist But Are Not Wired

These modules exist as files in the codebase but are never invoked by the discovery pipeline. They are listed here so the gap between what is built and what runs is visible rather than silent.

twitter-scraperjina-reader-scraper

Stating the boundary is what makes an intelligence product credible. A coverage page that only lists what exists is marketing; the notCovered list above is the most important part of this page. Every entry is grounded in a specific behaviour of the code — a missing integration, a deliberate design choice, a structural limitation — rather than a vague statement of future intent.

New to The Tell? Take a guided tour tailored to your role.