Tags: Tracer-Cloud/opensre
Tags
chore(deps): bump anyio in the uv group across 1 directory (#6321) Bumps the uv group with 1 update in the / directory: [anyio](https://github.com/agronholm/anyio). Updates `anyio` from 4.13.0 to 4.14.2 - [Release notes](https://github.com/agronholm/anyio/releases) - [Commits](agronholm/anyio@4.13.0...4.14.2) --- updated-dependencies: - dependency-name: anyio dependency-version: 4.14.2 dependency-type: indirect dependency-group: uv ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
chore(deps): bump anyio in the uv group across 1 directory (#6321) Bumps the uv group with 1 update in the / directory: [anyio](https://github.com/agronholm/anyio). Updates `anyio` from 4.13.0 to 4.14.2 - [Release notes](https://github.com/agronholm/anyio/releases) - [Commits](agronholm/anyio@4.13.0...4.14.2) --- updated-dependencies: - dependency-name: anyio dependency-version: 4.14.2 dependency-type: indirect dependency-group: uv ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
chore(deps): bump anyio in the uv group across 1 directory (#6321) Bumps the uv group with 1 update in the / directory: [anyio](https://github.com/agronholm/anyio). Updates `anyio` from 4.13.0 to 4.14.2 - [Release notes](https://github.com/agronholm/anyio/releases) - [Commits](agronholm/anyio@4.13.0...4.14.2) --- updated-dependencies: - dependency-name: anyio dependency-version: 4.14.2 dependency-type: indirect dependency-group: uv ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Tell new users to run opensre to get started instead of opensre setup. ( #6302) Account sign-in already happens on first launch of the shell, so the getting-started docs no longer send people through a separate setup command. Keep opensre setup --dev for local webapp work and opensre integrations setup for connecting tools.
fix(work-items): schedule reminders at exact requested time (#6225) * fix(work-items): schedule reminders at exact requested time * fix(work-items): reject nonexistent DST reminder times * fix(work-items): distinguish DST gap validation * test(ci-loop): avoid weekday letter assertion * refactor(work-items): simplify reminder validation * fix(scheduler): migrate legacy reminder fire times * fix(scheduler): isolate reminder migration failures * fixing codex review --------- Co-authored-by: OpenSRE Agent <opensreagent@opensre.com>
feat(analytics): expand product events for install, tools, Ask User, … …and workflows (#6261) * feat(analytics): capture install, tool, Ask User, and workflow product events Measure install conversion, tool outcomes, Ask User prompts, and OpenSRE-created commits without sending tool payloads, OAuth URLs, or unredacted installer env values. * Fix install retries and preserve Ask User selection metadata Persist the install marker after successful delivery so failed installer sends can retry with the same event ID. Pass picker indexes and separate custom answers to analytics instead of splitting joined labels. * epoch observer improvements * Publish deduplicated agent-fixed CI epochs from repair verification * fix(github): isolate epoch history to the target PR --------- Co-authored-by: davincios <jvf.hus@gmail.com>
docs(shell): document five missing slash commands and the picker's st… …ay-open menus (#6241) * docs(shell): document five missing slash commands and the picker's stay-open menus Diffing SLASH_COMMANDS against the reference found 56 registered against 47 documented. /demo, /exit, /quit and /memory are covered on other pages; /choose, /gateway, /posthog, /runbooks and /theme appeared nowhere in docs/. Each is written from the registry's own description, usage and notes. The TTY menu table listed /model's options without saying the menu stays open, so picking `show` and landing back on the menu reads as a bug. Both the table and the /model entry now say a read-only pick returns you to the menu and that done or Esc leaves it. /theme and /choose open pickers too and are added to that table. Fixes #6240 * docs(shell): scope the stay-open note to /model, name the /choose demo exception Greptile review on #6241. The note sat under the whole TTY menu table but only /model was verified to behave that way: /theme closes after one pick, and only privacy_cmds, settings_cmds, integrations and model actually loop. Telling readers a menu stays open when it has already closed is worse than saying nothing, so the note now speaks for /model alone. /choose off-TTY also has an exception the entry missed: a pending onboarding demo choice is dropped with "Guided demo selection is unavailable here" rather than listed as options to reply to.
fix(ci-fix): merge a behind base before fixing and recover from post-… …push conflicts (#6237) * remove delete repo requirement * fix(loops): do not list the opened run again under its own report /loops show --run prepended the requested attempt to the recent-history query, so an attempt that was already among the 20 most recent runs appeared once as the opened report and again in the Recent runs table. Drop it from the table by run id after legacy-report restoration. Co-authored-by: Cursor <cursoragent@cursor.com> * test(skills): pin loop-only demo cleanup and a non-dumb TERM for table geometry The scheduling card no longer deletes the demo repository, but its colocated test still asserted the `["repo", "delete"` call. Assert the new contract instead (loop removed, repository reported as remaining) and tidy the sentence fragment the removal left behind in Step 9. Rich renders "dumb" terminals at a fixed 80 columns regardless of the explicit Console width, so test_compact_table_limits_rows_and_prioritizes_findings[40] failed under TERM=dumb. Pin TERM in the test so the geometry assertions are deterministic. Co-authored-by: Cursor <cursoragent@cursor.com> * Refactor project structure and behavior * fix(ci-fix): merge a behind base before fixing and recover from post-push conflicts The repair only merged the base branch when GitHub already reported the PR as DIRTY. A PR that was merely behind (Tracer-Cloud/opensre-webapp#40, 38 commits behind main) was fixed on its stale head: the coding agent re-fixed a lockfile that main had already fixed, and that commit created the conflict GitHub then reported after the push. - Probe git for base commits the PR head lacks and merge them in first, so the fix lands on the current base and a failure already fixed there needs no change; the coding task says so. - When post-push verification reports the head as conflicted, merge the base into the pushed head once, push, and re-verify instead of stopping. - Blocked recovery keeps the pushed fix commit visible in the reply. Also repair two stale tests: fix_github_pr_ci returns a copy with work_outcome attached (identity assertion), and the schedule-line check tripped on Tue/Thu (asserted "T" not in the weekday). Co-authored-by: Cursor <cursoragent@cursor.com> * fix(ci-fix): never resume a settled repair; hold the demo lock as an OS file lock Review follow-ups on #6237. - resume: a prepared push whose checks already passed (or failed, conflicted, superseded) stays in the ledger, and the resume path accepted it whenever the PR head still matched. Every later scheduled tick then re-verified the same commit and reported another successful repair instead of "no failing checks". Only an unverified or timed-out record may resume. - demo lock: the lock was a bare directory removed by the helper's own finally block, so a helper killed on cancellation or timeout left it behind and every retry failed with "An attempt holds the demo lock". The lock is now flock/msvcrt on a lock file, which the OS releases with the process. Co-authored-by: Cursor <cursoragent@cursor.com> * fix(ci-fix): scope-check the resolver's edits inside a base merge commit With allowed_paths set, the scope check measured committed changes from the merge commit, so files the coding agent edited while resolving conflicts (including a weakened test) were never inspected, and the post-push conflict recovery merged and pushed with no scope check at all. merge_commit_edits returns the paths a merge commit changed relative to both parents - the resolver's hand edits, not the base's or the head's own changes. _enforce_scope adds them to the checked set and runs before both pushes. Co-authored-by: Cursor <cursoragent@cursor.com> --------- Co-authored-by: Cursor <cursoragent@cursor.com>
PreviousNext