agentgui 1.0.1085 → 1.0.1086
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/.gm/prd.yml +12 -0
- package/package.json +1 -1
package/.gm/prd.yml
CHANGED
|
@@ -3137,3 +3137,15 @@
|
|
|
3137
3137
|
subject: test.js lint-tokens.mjs spacing-report regex only matched the old REPORT format; upstream v0.0.349 changed lint-spacing output to PASS/baseline format, silently breaking the test's ability to catch regressions until this run's test execution surfaced it
|
|
3138
3138
|
witness: bun test.js failed with 'lint-spacing report line not found' after rebasing onto origin/main (8436f0a); fixed by widening the regex to match both formats
|
|
3139
3139
|
status: completed
|
|
3140
|
+
- id: kit-withbusy-double-fire-guard
|
|
3141
|
+
subject: Port docstudio's withButtonBusy async-button double-fire guard as withBusy() (48th run, direct-implementation due to gm-plugkit outage)
|
|
3142
|
+
notes: 'docstudio utils/dom-busy.js:5-20 withButtonBusy disables the triggering button + aria-busy + label swap + guaranteed restore via try/finally, dropping re-entry while busy. agentgui had zero button.disabled guards anywhere in app.js. Ported as withBusy() in overlay-primitives.js (raw-DOM utility alongside trapTab/useLongPress), wired onto ACP restart, agent re-check, and both agent-list retry buttons.'
|
|
3143
|
+
status: completed
|
|
3144
|
+
witness: 'Kit built+tested (bun test.js all pass, all lints pass, docs regenerated 0 drift), rebased onto v0.0.354/v0.0.355 cleanly, pushed c7da8c0->bee6ab7, kit CI (Publish/Deploy-GH-Pages/ci) all green. agentgui: browser-witnessed via agent-browser CLI (independent of the broken gm-plugkit spool) - withBusy() live-executed against a real button showed calls:1 for two concurrent invocations (second correctly dropped), disabled:true/aria-busy:true/label swap mid-flight, full restore after; all 4 call-site wirings confirmed present via regex against the live-served /gm/js/app.js source. agentgui pushed 6c8c6bc->ea402c1, all 4 CI checks green.'
|
|
3145
|
+
- id: gm-plugkit-wasm-bootstrap-broken
|
|
3146
|
+
subject: gm-plugkit spool watcher cannot boot - upstream npm packaging defect
|
|
3147
|
+
notes: 'Verified via `npm pack gm-plugkit@2.0.1990` (clean tarball inspection): the published npm package genuinely does not contain a wrapper/ subdirectory (wasi-shim.js, fs-atomic.js, kv-store.js, task-manager.js) that plugkit-wasm-wrapper.js imports from at module load. Reproduced identically across @latest (2.0.1990) and pinned versions 2.0.1988/2.0.1985/2.0.1980/2.0.1975, with fully cleared bun pm cache and rm -rf of /config/.gmweb/cache/plugkit and /config/.gm-tools between attempts - not a local cache/stale-install issue. A separate, distinct failure also seen on retry: plugkit.wasm itself throws a WASI LinkError (`import function env:host_cwd must be callable`) on instantiation even when the wrapper JS modules are manually patched in from a bunx-cached copy. This session worked around the outage entirely: docstudio-cue research via Explore agent (no spool dependency), implementation via direct Read/Edit, live witnessing via the independent agent-browser CLI instead of the gm browser verb, git operations via git_push/git_finalize equivalents run directly through the git CLI (git_status before any push, per the raw-Bash-git-fallback discipline) since git_finalize itself is a spool verb.'
|
|
3148
|
+
status: pending
|
|
3149
|
+
blockedBy:
|
|
3150
|
+
- external
|
|
3151
|
+
witness: 'npm pack gm-plugkit@2.0.1990 && tar tzf gm-plugkit-2.0.1990.tgz confirms no wrapper/ entries in the published tarball at all, despite plugkit-wasm-wrapper.js:20,31-33 importing 4 files from that path. Not fixable from this repo - the fix belongs in the gm-plugkit npm package itself.'
|