Repository navigation
Apra Fleet v0.4.3 -- now available #552
kumaakh
announced in
Announcements
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Before you upgrade: 0.4.3 carries required upgrades and deprecations -- read Apra Fleet 0.4.3 -- deprecations and required upgrades (#496) first, especially if you use GitHub App mode or run fleet-sprint.
apra-fleet 0.4.3 focuses on dispatch reliability (especially on Windows), VCS auth that heals instead of flapping, safer/cheaper Beads syncs, Antigravity (AGY) running fleet-sprint roles end to end, and
execute_promptsession forking.Dispatch reliability, especially on Windows
A dispatch to a Windows member no longer hangs after the LLM process has exited while a leftover child (sandbox server, test run, filesystem crawl) still holds the output pipe -- dispatches now complete on process exit with a short drain window. Test runners are bounded by wall-clock time with a process-tree kill, the stall detector catches frozen log tails with no parseable timestamp and uses an adaptive cadence, and a leaked per-member busy lock (an SSH drop during pre-exec steps left the member "busy" until restart) is now released. Writing the Claude workspace-trust file no longer fails with
ENAMETOOLONGon Windows.VCS auth that heals instead of flapping
Credential cleanup now revokes the exact credential a deploy wrote, instead of unsetting the right git-config key while orphaning the real token file. A push rejected because a member's Dolt clone was stale is no longer misread as a credential failure, and
provision_vcs_authstopped clobbering its own registration.register_membernow detects and self-heals a member's VCS provider up front instead of failing hours into an unattended run, andprovision_vcs_authfailures now carry their human-readable cause instructuredContent. Azure DevOps gained support for legacy*.visualstudio.comhosts, and PR descriptions are deterministically capped at 3500 chars so Azure DevOps no longer rejects them.push+prtokens now carryactions:write, so an agent can re-run failed CI jobs on its own PR, and a new docs/github-app-setup.md documents exactly which App permissions each access level needs -- see #496 for the required App permission changes.Beads: safer and cheaper
Every sprint verifies which beads database each member is mutating, and the supervisor dashboard shows it -- a member whose workFolder isn't a git clone with the right
origin/sync.remotenow fails the precondition instead of silently mutating the wrong project's beads. Beads/Dolt syncs per sprint are fewer and cheaper (bulk scope-guard fetch, probe memoization, time-boxed retries, remote-tip fingerprinting). Child-bead creation now refuses an id collision instead ofbdsilently overwriting an existing bead. A new maintenance playbook documentsbd prune/bd flattenfor reclaiming Dolt storage -- read its prerequisites first: flattening rewrites history and every other clone must re-bootstrap.Antigravity (AGY) end to end
AGY members now run fleet-sprint roles: Claude-only tools with no Antigravity mapping are dropped from transformed agents (Antigravity CLI refused every role dispatch over
ToolSearch), tool-discovery prose in the 11 role prompts is gated per provider, and permissions are delivered into isolated per-project configs so members run under--mode accept-editswith exact per-tool allowlists rather than machine-global grants.execute_promptsession forkingNew
forkparameter (boolean or session id) branches a new session from an existing one -- implemented for Claude via--resume <src> --fork-session. Non-fork-capable providers returnfork_unsupported; an unknown explicit fork source is a terminalsession_not_found. Separately, every fleet-sprint dispatch site now statesresumeexplicitly, so one-off prompts no longer silently reattach to a member's last (possibly huge) session.Install hardening and release validation
The installer now fails loudly, before making any changes, when Node.js 22.16+/npm (required for fleet-se) are missing instead of silently skipping the beads step; on Windows,
install --forcereuses an existing elevated scheduled task instead of failing with Access Denied, and never reports success while the server is left stopped. A new repeatable fresh-install test harness (Windows Sandbox + a clean Docker container) exercises a no-Node install, a full install, and both upgrade paths before each tag.Also in this release
memory-contract/v1-- a machine-checkable spec, JSON schemas, MCP/OpenAPI bindings and a conformance suite for the 23kb_*/code_*tools; config-driven HTTP KB provider selection restored; a title-keyed three-way merge for.fleet/kb-canonical.jsonso harvest conflicts no longer force a blind ours/theirs pick.{{secure.NAME}}deprecated in favor of{{secret.NAME}}(see Apra Fleet 0.4.3 -- deprecations and required upgrades #496 for the migration and the breaking reason-code rename).better-sqlite3removed as a dependency (the knowledge bank uses builtinnode:sqlite); no data migration needed.Upgrade
or grab the latest installer from the release page and run
install --force.Read #496 before upgrading if you use GitHub App mode, run fleet-sprint, or reference
{{secure.NAME}}anywhere.Full release notes: https://github.com/Apra-Labs/apra-fleet/releases/tag/v0.4.3
All reactions