Tags: getmissionctrl/aweb
Tags
Merge aaed: server-version-bump guard (CI change-base + release floor… …) + guard-mandated 1.26.24 bump (reviewer-ACKed eed7b82) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
release: server v1.26.21 — scope-aware suggest-alias-prefix(scope,exc… …lude,count) endpoint for AC onboarding name+scope
release: aweb server 1.26.20 — m3.2 app event channel (/v1/events/app… … + subscriptions + SSE) + app emit-key signing (SoT 6.7)
release: aweb 1.26.19 (cli: default API timeout 20s -> 30s for hackat… …hon) Bump server 1.26.18 -> 1.26.19 at e5adf5e. CLI-only behavior change; server unchanged on the wire. Scope (cli/go/awid/client.go +8/-8 + http_transport_test.go +7/-7 + docs/configuration.md +5/-5, total +20/-20): - DefaultTimeout const raised from 20s to 30s for the awid client HTTP transport in the aw CLI. - Timeout assertion tests updated to match new default. - docs/configuration.md timeout reference updated. - SSE: no overall timeout (unchanged). - Write requests: no retries (unchanged). - Mutating-request timeout warning (unchanged). Validation chain under standing policy awebai#17 (Juan: never ship with failing tests, ever): - Grace pre-handoff: go test ./awid -run 'TestAPITimeout|TestNewAPITransport|TestMutatingRequestTimeout' PASS; go test ./cmd/aw -run 'TestConfigureBaseURLFallback|Test.*Transport|TestA2A' PASS; git diff --check clean. - Hestia make ship full chain at e5adf5e + 1.26.19 bump: ALL PASSED across server pytest (619) + awid pytest (218) + cli go test + channel tests (105) + federation e2e (116) + OSS user journey (455). Zero FAIL labels anywhere in output. Hackathon urgency: npm @awebai/aw 1.26.19 needed live for hackathon agents tonight; release-aw GHA + downstream awebai/aw goreleaser will publish to npm immediately. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
release: aweb 1.26.18 (E2EE chat journey blocker fix) Bump server 1.26.17 -> 1.26.18 at 96e7067. Scope: fix shared get_pending_conversations to match chat_participants by exact DID OR participant_agent_id (was exact DID only). CLI-origin chat sessions store hosted custodial peers under did_aw/stable id, while AC dashboard queries with hosted did:key + participant_agent_id; the participant-id check was ignored in the participant join, so CLI -> hosted custodial chat sessions were invisible in pending. Fix matches exact DID or participant_agent_id, prefers exact DID, and uses matched p.did for read receipts / unread / waiting logic. This closes the 2 E2EE journey failures that had been stable across v0.5.69 / .70 / .71 / .72 ships and held v0.5.73 in incident posture under release policy awebai#17 (Juan 2026-06-12, 'we cannot ship with failing tests, ever') + awebai#18 (Sofia 2026-06-12, 'identical labels twice = consistent broken, not flake'). Validation chain: - Olivia + Mia: APPROVED (aweb-aaqu engineering review). - Grace: AC make test-cloud-user-journeys-local-aw 308/308 with this aweb source (was 306/2 prior). - Server focused: test_chat_service + test_chat_http 83 passed; test_mcp_mail_chat 45 passed. - Hestia local make ship (96e7067 + 1.26.18 bump): full chain PASSED end-to-end. OSS user journey 455 PASSED. Server 619 + awid 218 + cli go + channel 105 + federation e2e + journey 455 = ZERO failing tests. Olivia's regression test_chat_service.py +32 lines exercises pre-fix-failure / post-fix-pass on the journey shape per reproducer-as-gate (awebai#12). Follow-up aweb-aaqv (same-class session open/reply/stream exact-DID guards) filed non-blocking by Olivia/Mia. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
release: aweb 1.26.17 (team-auth envelope v2 verifier) Bump server 1.26.16 -> 1.26.17 at 7473826 (fix-forward of 0bf8d3d). CLI version follows per Makefile coupling (#219 stays open). Scope: - shared Python verifier/spec in aweb.team_auth_envelope - Go aw id request --team-auth emits v2 signed envelope - Go/Python conformance vectors for team-auth-envelope-v2 - docs/team-auth-envelope-v2.md plus SOT updates documenting request binding and the replay-within-skew limitation - legacy compact v1 path stays independent of public-origin settings (fix-forward from 0bf8d3d regression caught by full make ship) Validation chain: - Grace pre-handoff at 7473826: targeted auth 36 passed, previously failing HTTP group 219 passed, full make ship PASSED end-to-end (server 619 + awid 218 + cli go + channel 105 + federation 116 + OSS journey 455). - Hestia local make ship at 7473826 + 1.26.17 bump: PASSED end-to-end (OSS journey 455, server checks green, channel green, federation e2e green). - Peer reviews on 7473826: Olivia APPROVED, Mia APPROVED, Athena protocol APPROVED, Rose APPROVED (Juan-mandated dual-review after the 0bf8d3d gate failure). - v2 audience acceptance is app.state.public_origin when set else settings public_origin (not a union); intentional narrowing per Olivia's review. AC mounted real-HTTP test will pin the external origin downstream. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
release: aw 1.26.16 (aweb-aaqm venue WiFi/NAT hardening) Carries be6db09 (HTTP client hardening for hostile venue networks) + acda65f (review-blocker folds: no write replay in baseURL fallback, header timeout follows AWEB_HTTP_TIMEOUT). Mia: full module go test + go vet clean. Grace independently approved at acda65f. Athena: no further blocker. Hackathon P0 — Juan wants this installed at the venue. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
release: aw 1.26.14 (aweb-aaqi aw init blockers) Three bugs fixed in d0dcb08 + 70c2395 (Olivia + Mia + Athena clearance): - repo_origin no longer sent on /v1/connect; canonicalize_git_url now accepts SSH-alias remotes for the surviving repos lane. - connect failures roll back local .aw state while preserving resumable partial-init. - already-registered names fail deterministically with both DIDs + recovery guidance, no mismatched key left behind. Unblocks aaqe.7 identity-creation flow (pi.aweb.ai/ama greeter provisioning). Conformance vector synced alongside (prior commit). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
PreviousNext