Skip to content

AI Engine Room 0.1.1

Latest

Choose a tag to compare

@gregweir gregweir released this 04 Sep 15:26
4f4eef7

AI Engine Room 0.1.1

AI Engine Room 0.1.1 is the first stable release of the deliberately narrow
Observe → Explain → Diagnose → Report desktop workflow.

What is included

  • Explicit, on-demand observation of supported same-machine Ollama and LM
    Studio endpoints and bounded machine context.
  • Passive traditional single-model llama.cpp detection on the tested Ubuntu
    baseline.
  • Optional one-run-authorized synthetic inference observations for eligible
    Ollama and LM Studio models.
  • Session-only observation history and deterministic diagnostic explanations.
  • An allow-listed report that can be explicitly copied or saved as UTF-8 plain
    text. Saving never replaces an existing file.

Supported baselines

  • Ubuntu 24.04 LTS x86_64 using the .deb package.
  • Windows 11 25H2 build 26200.7462 x64 using the NSIS installer.

Other operating-system versions, architectures, provider configurations, and
macOS are not currently claimed. See the
support matrix.

Verified downloads

Platform Exact hosted filename Bytes SHA-256
Ubuntu AI.Engine.Room_0.1.1_amd64.deb 4,896,518 c73dac2320632bbb6b4a3f02c65943a95bf40cefe83c367e8e65b5b7136c617a
Windows AI.Engine.Room_0.1.1_x64-setup.exe 2,762,354 e51fd579a5045599d99596f8b4bdfd10d3712b91b9110ed02fe72b3bc786dfb4

Use the SHA256SUMS.txt attached to this GitHub release as the checksum
manifest for these binary assets.

Both packages were built from source commit
4f4eef7ee05b12e5e3c6c6aaed162d13b5b654c7, tree
d72b25ae3f118400b0260c1d5d6cf9e0b5b9f990. Each exact package passed one
bounded physical acceptance run on its supported baseline, covering native
launch, core navigation, Refresh, Report, Copy, new-file Save, and existing-file
no-clobber behavior. See the
physical acceptance record.

Unsigned packages

The 0.1.1 packages are not digitally signed because the project does not
currently have an affordable signing route. Download only from the official
GitHub release, download SHA256SUMS.txt from the same page, and compare the
complete filename, byte size, and package SHA-256 before installation. Matching
details establish that the downloaded bytes match the published artifact; they
do not authenticate Tartanleaf.com Inc. as a cryptographic publisher.

Do not disable SmartScreen, antivirus, or another security control; install a
trust certificate; or bypass organizational policy. Continue only through a
normal operating-system option after verification and only when local policy
allows it. Exact commands and the bounded Windows continuation path are in the
user guide.

Deliberate limits

This release has no account, telemetry, automatic updater, background polling,
persistent history, provider management, network scanning, benchmark score,
model-fit recommendation, or broad platform claim. It is not privacy- or
security-certified.