Repository navigation
v0.16.16 — machine check, your own speech server
New
- Machine check. Settings → «Machine check» lists only what is not in order on the machine, with the command that fixes it.
- Your own speech server. Voice input can use an OpenAI-compatible transcription server of your own; its key is kept on the machine (mode 600) and is sent only to the address it was saved for — changing the URL needs the key again.
Install
Docker: docker pull ghcr.io/ly-solutions/harnessly:0.16.16 (the image is latest too, and with-agents
adds Claude Code and Codex). Binaries: take the archive of your system below, harnessly_0.16.16_<os>_<arch>.tar.gz
(it holds harnessly, runtime and agent). How to run, verify and update:
docs/deploy.md.
Changelog
Verify
cosign verify-blob checksums.txt --bundle checksums.txt.sigstore.json \
--certificate-identity "https://github.com/ly-solutions/harnessly/.github/workflows/release.yml@refs/tags/v0.16.16" \
--certificate-oidc-issuer https://token.actions.githubusercontent.com
sha256sum --ignore-missing -c checksums.txt
gh attestation verify harnessly_0.16.16_linux_amd64.tar.gz --repo ly-solutions/harnesslyAll changes, including those the notes leave out: CHANGELOG.md.