v1.3.2

What's Changed

• feat(release): /publish refuses to tag a main commit whose ci run is red, pending or missing by @using-system in #54
• fix(actions): retry the run-time downloads on a connection reset, not only on a timeout or a 5xx by @using-system in #56

Full Changelog: v1.3.1...v1.3.2

v1.3.1

What's Changed

• fix(ci): pin the workflows' Ubuntu runners to ubuntu-26.04 - the ubuntu-latest label migrates to Ubuntu 26.04 from 2026-10-19 by @using-system in #36
• refactor(odd-status): one launch script per CLI at the repository root, generic over the packaged command by @using-system in #39
• fix(actions): list the characters resolve-oddyssey.sh accepts instead of a locale-dependent range by @using-system in #40
• docs(agents): run a security review of the diff alongside the review subagent on every PR by @using-system in #47
• fix(ci): release.yml refuses a version below the highest release and a major tag move that is not a fast-forward by @using-system in #48
• fix(odd-status): keep the job token out of the checkout and say what --secret-env-vars does by @using-system in #49
• fix(actions): pin the uv release every setup-uv step installs and disclose it by @using-system in #50
• fix(odd-status): export the Claude credential from the shell instead of putting it on env's argument vector by @using-system in #51
• fix(odd-status): recompute the verdict with the package's own script on the checkout instead of reading it from the model's answer by @using-system in #52

Full Changelog: v1.3.0...v1.3.1

v1.3.0

What's Changed

• feat(actions): one minimum oddyssey version, checked by every setup, and odd-status reads the package's verdict only by @using-system in #34

Full Changelog: v1.2.4...v1.3.0

v1.2.4

What's Changed

• refactor(odd-status): read the verdict the status emits - the model writes the summary, never the status by @using-system in #32

Full Changelog: v1.2.3...v1.2.4

v1.2.3

What's Changed

• fix(odd-status): a skipped ledger row is a fact the memory invariant reports, never a violation by @using-system in #29

Full Changelog: v1.2.2...v1.2.3

v1.2.2

What's Changed

• fix(odd-status): the loop state's Action column is the verdict's anchor - gaps and deferrals are facts for the summary by @using-system in #27

Full Changelog: v1.2.1...v1.2.2

v1.2.1

What's Changed

• fix(odd-status): what the memory records as settled is a fact, never a warning - by-design gaps, declined findings, deferrals no memory write can close by @using-system in #25

Full Changelog: v1.2.0...v1.2.1

v1.2.0

What's Changed

• chore(release): add a /publish command that cuts a release from a Claude Code session, the way oddyssey's /oddyssey-publish does by @using-system in #13
• feat(odd-status): read github.token itself for the Copilot run - no GITHUB_TOKEN line on the caller's step by @using-system in #15
• docs(readme): say the actions bring oddyssey's capabilities into a workflow, not that they run prompts headlessly by @using-system in #17
• refactor(actions): every step's logic as a script under /scripts/, the action.yml as the wiring only by @using-system in #20
• ci(tests): one tests// tree for every action, run by one matrix job - each script off the runner with fake tools, then the action for real and its runner tests by @using-system in #21
• feat(setup-claude): install Claude Code and the oddyssey package on a runner, an Anthropic API key or a Claude OAuth token as its credential - and odd-status runs through it by @using-system in #23

Full Changelog: v1.1.0...v1.2.0

v1.1.0

What's Changed

• fix(setup-copilot): authenticate the headless run with the workflow's GITHUB_TOKEN, no personal token secret by @using-system in #3
• feat(setup-copilot): default the model input to gpt-5.6-luna, the cheapest model the benchmark measured by @using-system in #5
• fix(ci): close the security audit - pinned checksum-verified CLI download, no token to third-party code, validated inputs, scoped headless run, release cut from main only, Security section in AGENTS.md by @using-system in #7
• feat(setup-opencode): install opencode and the oddyssey package on a runner, an OpenAI-compatible endpoint as its provider - and both setups export the CLI and the model by @using-system in #9
• feat(odd-status): run the packaged /odd-status through the CLI a setup action installed, and turn its verdict into outputs a workflow can gate on by @using-system in #11

Full Changelog: v1.0.0...v1.1.0

v1.0.0

What's Changed

• feat(setup-copilot): install the Copilot CLI and the oddyssey package on a runner, the model and the package version as its only inputs by @using-system in #1

New Contributors

• @using-system made their first contribution in #1

Full Changelog: https://github.com/using-system/oddyssey-actions/commits/v1.0.0