Tags: paulirotta/ahma
Tags
fix(ahma_mcp): mark the last AppContainer-dependent Windows test #[ig… …nore] (#559) reads_outside_the_scope_are_blocked only passed on windows-latest CI by accident: the (now-disabled, see #558) AppContainer grant denied every read, in-scope included, so a read genuinely outside the scope also failed and satisfied the assertion for the wrong reason. With Windows spawns now falling back to Job-Object-only containment (no path confinement per SPEC R6.3.9), the read legitimately succeeds and the test fails honestly instead of passing by coincidence. Ignored with the same rationale as the other two appcontainer gate tests, and SPEC R6.3.3's status note updated to cover all three.
fix(ahma_mcp): drop the hardcoded path separator in expand_home tests (… …#554) `test_expand_home` asserted `expanded.ends_with("/test")`. #542 routed this module's `expand_home` through `ahma_common::config::expand_home`, which joins with `PathBuf::push` instead of the previous `replacen("~", home, 1)`. On Windows that changed the result from `C:\Users\x/test` (mixed separators, the literal `/` preserved from the input) to `C:\Users\x\test`, so the assertion failed. The new behaviour is the correct one — a canonical platform separator rather than a mixed-separator path — so the test is what was wrong. It is exactly the assumption AGENTS.md rules out: "Use `MAIN_SEPARATOR`/`Path` APIs, never string separator assumptions." Both this test and its `~\` sibling now compare against `std::path::MAIN_SEPARATOR`. This was the sole failure in the post-merge Windows matrix for 0.19.1 (4352/4353 passed), and it cascaded through `CI green` to skip `Publish Release`, so v0.19.1 was never published. The release gate keys on whether the tag exists rather than on the version changing, so landing this re-triggers the 0.19.1 build and publish without a further bump.
chore(release): bump version to 0.16.7 (#513) Ships /skill support for the Agent Skills open standard in the TUI (#512), honest-panes disclosure work (#511), window process-tree reaping and detail overlay fixes (#510), bridge shutdown reporting (#507), and the await-timeout and tool-list unification work (#503–#506).
[ci] Release 0.16.6 — relax build-time macOS signing to unblock insta… …ll (#498) * ci: temporarily drop build-time macOS codesign step (relax R-SIGN.1) Reverts #493 (fdb0cec). v0.16.5 is the first release to ship a codesign'd darwin-arm64 binary and it fails to install on macOS. Until the Apple Developer signing path is finished, ship the plain linker binary again — install.sh / 'ahma update' still retrofit the local hardened-runtime re-sign, matching the pre-0.16.5 posture that installed cleanly. The build-time signing step returns as an open (unlanded) PR so the work is not lost. * chore(release): bump version to 0.16.6
PreviousNext