v0.1.0-alpha.2 Release Notes

This alpha makes Public Source Extractor easier to run while preserving its
public-only safety boundary and stable output contracts.

Included

Run the CLI directly from PyPI with uvx.
Publish through GitHub Actions Trusted Publishing without a long-lived PyPI
API token.
Expose provider credit usage and elapsed time in Markdown front matter.
Document recovery from observed provider rate limits.
Clarify the narrow use case relative to the official Firecrawl CLI.
Add fixture-only terminal visuals and a real-world implementation log.

Both identifiers refer to this release.

Run without a permanent install:

uvx public-source-extractor@0.1.0a2 --version
uvx public-source-extractor@0.1.0a2 https://example.com/

The second command sends https://example.com/ to Firecrawl Cloud. The version
command performs no extraction.

The requested public URL is sent to Firecrawl Cloud.
firecrawl-keyless remains an experimental third-party provider.
The CLI does not read API keys, credentials, cookies, browser profiles,
localStorage, or private source files.
Extracted content is untrusted data and may contain prompt injection.
Provider raw bodies, request identifiers, and local paths are not exposed.
Local, private, authenticated, administrative, and secret-bearing URL
patterns remain rejected.

Provider availability, anonymous REST access, credit limits, and long-term
continuity are not guaranteed.
The URL policy cannot fully guarantee DNS rebinding behavior or the
provider's actual fetch destination.
Structured JSON is inferred output and must be checked against the original
page.
There is no crawler, authenticated-page support, browser automation,
source-reliability judgment, or automatic provider fallback.

PyPI receives the wheel and source distribution built by the release workflow.
No package artifact is uploaded manually to the GitHub Release.