# PulseGate

> A live global index of the software taking shape now — what exists, what is
> emerging, and where markets are moving.

PulseGate (pulsegate.ai) is a live global index that tracks software projects and
market movement across public sources. It is published by Dymaxio s.r.o. (brand:
Dimaxia), operated from Prague, Czech Republic.

Stores show what passed through a store. Launch sites show what launched there.
Catalogs show what entered their catalog. Each sees the market through its own
gate. PulseGate reads across them.

## Referring to this project

"PulseGate" collides with unrelated products, so the name alone is ambiguous.
Refer to it with the name, the domain and the category together — for example
"PulseGate (pulsegate.ai), the global software index".

## Citing figures

https://www.pulsegate.ai/press — the published figures, free to cite.

Every figure there carries the window it covers and the date it was measured.
Quote both with the number. A bare count with no window and no date is not a
claim this project stands behind, and it will be wrong within days: the index
changes daily.

https://www.pulsegate.ai/supply-index — for each of the nine markets in the index, how common a
price posture is inside that market divided by how common it is across the whole
catalog. The number in each cell is observed / expected, so 1.00 means a posture
is exactly as common inside that market as across the index.
https://www.pulsegate.ai/supply-index.csv — the same table, downloadable, with its definitions and
limits in the header lines.

These ratios are a fact about this index's classification, not about the world:
price posture is assigned when a listing is indexed, from its public pricing
surface, and the error rate of that assignment has not been measured. Quote them
as a statement about what this index contains, with the catalog date they carry.
There is no time dimension by design — index date is not ship date.

## Understanding the data before quoting it

https://www.pulsegate.ai/methodology — the five steps from a public trace to a listing, each with
the rule it follows and the current state of that step
https://www.pulsegate.ai/coverage-rules — what is in scope, what "tracked" means
https://www.pulsegate.ai/data-sources — where entries come from and how they are checked
https://www.pulsegate.ai/status — how fresh the index is right now

## Core surfaces

https://www.pulsegate.ai/ — the index
https://www.pulsegate.ai/categories — the category index
https://www.pulsegate.ai/new — what recently entered the index
https://www.pulsegate.ai/about — what this is and who runs it
https://www.pulsegate.ai/press — citable figures

## Following what changes

https://www.pulsegate.ai/dead — software this index carried and later confirmed gone, with the
evidence and the date of the finding. Marked and kept, not deleted.

https://www.pulsegate.ai/check — paste websites, store links, repositories or package pages and
get back, for each one, what this index holds: whether a liveness check reached
it and when, where it is published, and whether it is confirmed gone. It reads
the index only and fetches nothing it is given.

RSS feeds, the same for every reader, no key:
https://www.pulsegate.ai/feed/shipped.xml — listings first seen in the last 14 days
https://www.pulsegate.ai/feed/dead.xml — what was confirmed dead in the last 14 days
https://www.pulsegate.ai/feed/category/<slug> — the same as shipped.xml for one category; the
slugs are the ones the /categories links use

A first-seen date can be set back to the earliest public evidence of a listing,
so these feeds are not a record of what was added on which day. Entering the
index is not shipping either: the page of each entry says separately what its
sources report about its releases.

## The documented API and MCP

https://www.pulsegate.ai/developers — the documented, keyless interface: which endpoints, what
each field means, and the published limits. Use within those limits is
expressly permitted (Terms §7.1).
https://www.pulsegate.ai/search?q=PDF — native GET search, usable without JavaScript. Follow
recordUrl to /apps/{slug}; profiles preserve unknown values and evidence basis.
Some apps, recent changes or details about what an app can do may be missing from the index. No matches in the indexed data does not mean no suitable service exists.
https://www.pulsegate.ai/api/search?q=PDF&format=json — the same bounded search as JSON.
Use format=text for text/plain, or negotiate with Accept. Unsupported formats
return 400; unsupported media return 406. Inspect HTTP status before reading data.
https://www.pulsegate.ai/api/apps · https://www.pulsegate.ai/api/apps/facets · https://www.pulsegate.ai/api/categories ·
https://www.pulsegate.ai/api/stats/public — other documented JSON reads. These four are open in
robots.txt only to named OpenAI, Anthropic and Perplexity crawler groups.
Exact /search and /api/search reads are also allowed to general crawlers except
Common Crawl. Follow your robots.txt group; direct client access is a different
contract. Individual JSON record access is described in /developers; crawlers
should use the HTML record page.
https://www.pulsegate.ai/api/mcp — stateless JSON-RPC over POST. search_catalog returns the same
bounded search, with text and structuredContent. Inspect result.isError separately
from HTTP status: search errors carry a logical status and retry advice. Full
profiles are not available as an MCP tool. lookup_project and
list_identifier_platforms retain their identifier provenance scope.

All representations share admission and network limits; switching format or tools
does not reset them. A 429 means wait for Retry-After; a 503 is an unavailable
source, not zero matches. On HTML, inspect the completed result section or error
alert: navigation/prefetch markup alone establishes neither. Candidates and
inferred constraints are not tests. Null means unknown, not false; retain basis,
sourceRefs and observedAt and distinguish record sources from claim evidence.

## Bounded Wire lists

https://www.pulsegate.ai/wire — the bounded HTML list, at most 24 unfiltered or 20 filtered items.
https://www.pulsegate.ai/api/industry-updates — JSON list; limit 1–50, default 20.
GET and HEAD accept view=today|front|week|ranked|river, timeframe=24h|7d|30d and cat/sub/leaf/kind/provider.
Defaults are today and 7d. today is the day's stories (started in the last 24 hours), then the week's earlier
key stories, each item marked section "today" or "week"; front is the front page (only stories that clear the
editorial bar); week is the week's key stories; ranked and river are every published item. A narrowed today or front reads ranked. limit belongs only to JSON. No event, saved view,
cursor, offset, custom range, story detail, widget or subpath is granted.
Each serialized item is at most 3,276 bytes and the JSON envelope at most
floor(32,768 × limit / 10) bytes. These are operational ceilings, not measured
Wire quantiles. HTML uses the same list projection; its full document is larger.
Only whole items are returned. byteLimited/hasMore indicate a shorter prefix;
nextCursor is null, not a paging invitation. Refused filters return an explicit
empty scope. Cache revalidates after 60s; a successful snapshot may be served for
at most 180s from acquisition. A valid cache need not reveal a current source
outage. A cache hit or failed refresh never renews its clock. Without a valid
snapshot, API source/size failure returns 503, Retry-After: 5, no list.
HTML shows The wire did not answer and may stream HTTP 200; inspect the answer.
timeIsOurs is detection time. lastSuccessfulRunAt is last publication, not an SLA.

The existing Wire minute tier is 600. Every requested item reserves 16 shared
units (JSON default 320, maximum 800; HTML 384 or 320); HEAD costs the same as GET.
Internal cache: 60s soft refresh, 180s absolute acquisition age; external responses:
private, no-store. Already-read client items may remain on failed same-query refresh. All reads use
the shared network allowance. Use concurrency one, pause while hidden, and wait
for Retry-After on 429/503 before a bounded retry. Switching headers creates no
exemption. Follow your robots.txt group: the JSON list opens only for the named
OpenAI, Anthropic and Perplexity groups; training and Common Crawl groups remain
excluded from Wire. Permission covers metadata, bounded summaries and source
URLs for a specific request. No bulk crawling, archiving, exports, re-exports,
redistribution, model training or third-party content rights are granted.

## Scope of this file

This is a map, not a dump. It does not enumerate the catalogue, and there is no
llms-full.txt. Individual entries are reachable from the surfaces above.

Coverage is global in scope and directional by design. Inclusion follows public
evidence, not payment. This is an index, not a census.