Skip to content
Open-Source Registry

pypi / skills-analyzer / 0.1.1

skills-analyzer 0.1.1

bitbison: flagged, behavior flagged, dependencies clean, posture not checked, pipeline clean, 1 of 1 jobs captured RiskmediumVulnsnone
processes
410
operations
22K
packages
0
compiler runs
3
connections
2
build time
18s
job time
18s
capture
188 KB
records
34K
bytes / record
5.6
run c088db13-233b-48ab-a595-affe4fcc939a · built 2026-11-01 · complete View on PyPI
Reviewed by Bitbison: maliciousskills-analyzer 0.1.1 depends on is-it-ai==0.1.1 and calls its verify() from its own command. That code posts an environment and organization fingerprint to a Cloud Run endpoint and runs a file the server returns on Windows and macOS. The other releases are covers from the same publisher.

Graph

egress · medium
2

Checks

11 of 12 clear Network activity
Build behavior▲✓✓✓✓✓✓✓✓
Supply chain✓✓
Pipeline✓

Detections

1
  • Connected to a default hosting domain: cdn-9zp0pi12ck0dmfig-3ozmpyh7ja-uc.a.run.app:443
    cdn-9zp0pi12ck0dmfig-3ozmpyh7ja-uc.a.run.app:443 › cdn-9zp0pi12ck0dmfig-3ozmpyh7ja-uc.a.run.app:443
    medium

Detection

mediumwitnessed

Connected to a default hosting domain: cdn-9zp0pi12ck0dmfig-3ozmpyh7ja-uc.a.run.app:443

Unexpected network connection

Subject
cdn-9zp0pi12ck0dmfig-3ozmpyh7ja-uc.a.run.app:443
Destination
cdn-9zp0pi12ck0dmfig-3ozmpyh7ja-uc.a.run.app:443
Jobs
build-import
Rule
egress · skills-analyzer

Evidence

  1. Connection outside the process’s allowed destinationscause

    opened by skills-analyzer

    subject cdn-9zp0pi12ck0dmfig-3ozmpyh7ja-uc.a.run.app:443

    by /home/runner/work/skills-analyzer/skills-analyzer/.venv/bin/python /home/runner/work/skills-analyzer/skills-analyzer/.venv/bin/skills-analyzer

  2. What kind of service the host issupport

    Google Cloud Run default domains

What came before it

Loading the graph

Processes

build-import

python›python

Ran
/home/runner/work/skills-analyzer/skills-analyzer/.venv/bin/python /home/runner/work/skills-analyzer/skills-analyzer/.venv/bin/skills-analyzer
Invoked by
.venv/bin/python /home/runner/work/_temp/imports.py skills-analyzer
Lineage · 2 processes
  1. python.venv/bin/python /home/runner/work/_temp/imports.py skills-analyzer2729
  2. python/home/runner/work/skills-analyzer/skills-analyzer/.venv/bin/python /home/runner/work/skills-analyzer/skills-analyzer/.venv/bin/skills-analyzer2733

Remediation

Find the code that made the connection and what it sent. If the service is the project's own, accept it with an accepted_download row naming the program and the host.

Network

2 peers
  • cdn-9zp0pi12ck0dmfig-3ozmpyh7ja-uc.a.run.app:443 skills-analyzer's own build code · skills-analyzer · Google LLC undeclared
  • files.pythonhosted.org:8080 the package manager · python3.14 (skills-analyzer) via Package registries profile (pip → files.pythonhosted.org) ecosystem
Versions3RiskVulnsPublished
0.1.2did not build1d ago
0.1.1mediumnone1d ago
0.1.0nonenone3d ago

Raw data

capture
188 KB
records
33,613
bytes per record
5.6

Loading a sample of the capture

Want to see more? E-mail us at hello@bitbison.io about your use case.

1 of 1 jobs captured: build-import success · Generated 2026-11-01 · run c088db13-233b-48ab-a595-affe4fcc939a

analysis 0.1.0 · 2026-10-31 · 249a8cc · graph v2

skills-analyzer 0.1.1 pulls in is-it-ai 0.1.1, which sends a machine fingerprint out and runs the file it returns

maliciousCovers 0.1.0, 0.1.1 and 0.1.2Reviewed by BitbisonUpdated 2026-11-01

skills-analyzer 0.1.1 depends on is-it-ai==0.1.1 and calls its verify() from its own command. That code posts an environment and organization fingerprint to a Cloud Run endpoint and runs a file the server returns on Windows and macOS. The other releases are covers from the same publisher.

What it does

skills-analyzer 0.1.1 declares is-it-ai==0.1.1 as a dependency. Its skills-analyzer command calls verify() first thing after parsing its arguments, before any of its own work, on every run. Running the command with no arguments connects to the endpoint.

is-it-ai 0.1.1 presents itself as a library that detects AI agent environments. Its verify() collects the operating system, the Claude Code process and entrypoint variables, whether managed settings are present, and on Windows and macOS a fingerprint: hashes of a corporate MCP gateway certificate and of an MDM-assigned preferences file, which identify the organization a machine belongs to. It posts this to https://cdn-9zp0pi12ck0dmfig-3ozmpyh7ja-uc.a.run.app/v1/session/info. When the reply names a URL and a destination, it downloads that file; on Windows it opens it and on macOS it makes it executable and starts it detached. The endpoint and a list of AI tool, editor and model runner paths on Windows and macOS are bundled in is_it_ai/config/service.json.

The other releases of skills-analyzer do not pin is-it-ai; they are covers uploaded by the same publisher.

Timeline

  • First release on PyPI: 2026-10-28 22:24 UTC.
  • skills-analyzer 0.1.1, the release carrying the code, uploaded 2026-10-31 00:39 UTC.
  • Reported to the OpenSSF malicious packages database as MAL-2026-18427: 2026-10-31 12:55 UTC, 12 hours after that release was published.
  • Every release was still available on PyPI, none yanked, when this review was written.

Indicators

hosts
cdn-9zp0pi12ck0dmfig-3ozmpyh7ja-uc.a.run.app
requests
https://cdn-9zp0pi12ck0dmfig-3ozmpyh7ja-uc.a.run.app/v1/session/infohttps://cdn-9zp0pi12ck0dmfig-3ozmpyh7ja-uc.a.run.app/v1/status
packages
pypi:is-it-ai==0.1.1

Examined

0.1.1
sha256:1821b52563f74d83c3ecf8b73416a246a7384ab7fb716ce31e65a8fcc467a2f9

Links