2026.6.7 — Deterministic Gates, Docker Ready

Released 2026-06-07

← All releases

What's New

  • Deterministic workflow gate coverage added for end‑to‑end testing, ensuring gate logic is reliably verified.
  • Declarative tool‑review gates and multi‑agent review stages introduced, letting workflows enforce output policies automatically.
  • Docker‑based local development environment and optional Ollama stack now ship out‑of‑box, simplifying AI‑tool testing.
  • New workflow run action and public branding endpoint expose richer control and branding data to integrations.
  • Workspace script staging and output capture added, enabling scripts to run inside claws with result collection.

Improvements

  • Stream segment order preserved after refresh and pipeline‑done signals now prioritized for smoother UI updates.
  • Activity timeline pagination, history preservation, and visibility improvements boost user transparency.
  • Cross‑tab authentication now stores tokens in localStorage, enabling seamless multi‑tab sessions.
  • Workflow controls get pause buttons and visible progress updates, giving operators finer control.
  • Resilient Daytona workflow runs now survive workspace, policy, and model timeouts, reducing flaky executions.

Fixes

  • Fixed output_matches trigger re‑firing and ensured gates run even on non‑zero exits with valid JSON.
  • Resolved numerous greptile review feedback bugs: tool‑review gates, judge stage, PR validation, and placeholder routing.
  • Stabilized Docker provider, Ollama bootstrap, and external webhook creation to be atomic and error‑free.
  • Eliminated ghost split‑turn broadcasts, duplicate LLM turn sends, and cross‑tab auth race conditions.
  • Patched SSH host key handling, workspace route loops, and timeline pagination cursor issues.

Raw Changelog

Show raw git changelog
text
bb9de7d Add deterministic workflow gate E2E coverage (Marc Campbell)
b8a927c Preserve workflow gates and parse noisy outputs (Marc Campbell)
7040cd3 Address SSH workspace path review (Marc Campbell)
95d20dd Address workspace script staging review (Marc Campbell)
a8263d8 Stage workspace scripts in claws (Marc Campbell)
ff0fbf6 Merge main into done trigger branch (Marc Campbell)
b4699fd Merge pull request #354 from elasticclaw/codex/factory-pr-default (Marc Campbell)
03259f6 Merge pull request #364 from elasticclaw/feat/replicated-provider-contract (Marc Campbell)
44f019a Track pipeline done signals from websocket (Marc Campbell)
07638f4 Suppress ghost split-turn broadcasts (Marc Campbell)
58e54da Prioritize pipeline done triggers (Marc Campbell)
484f04e Preserve stream segment order after refresh (Marc Campbell)
38eb77d fix: prevent output_matches trigger from re-firing on every message (Claw)
d280e53 fix: normalize skipped verdict to pass for auto-transition when TreatSkippedAsPass is set (Claw)
592efa8 fix: evaluate gate even when run exits nonzero with valid JSON (Claw)
f12ef92 fix: wire output_matches trigger into checkPipelineMessageTriggers (Claw)
7139b79 fix: address greptile review comments on tool review gates (Claw)
bd84dbd feat: add declarative tool review gates for workflow run outputs (Claw)
e863f1a fix: address greptile review comments on judge stage (Claw)
a3d226e feat: add multi-agent workflow review stages (judge) (Claw)
007639f Defer replicated env injection during provision (Marc Campbell)
6739e25 fix(replicated): reject env in ProvisionClaw and correct docstring (Claw)
91760d7 feat(replicated): clarify file/env injection contract with explicit errors (Claw)
7b95971 Avoid double-decoding settings route parts (Marc Campbell)
9b73bb6 Fix settings workspace route loop (Marc Campbell)
7f1116c Address branding review feedback (Marc Campbell)
0a2a0ae Add public branding endpoint (Marc Campbell)
1cc7bca Fix timeline pagination cursor (Marc Campbell)
0ef616c Address activity summary review feedback (Marc Campbell)
7f5b3aa Cover activity-heavy timeline state (Marc Campbell)
8717f95 Preserve history with activity summaries (Marc Campbell)
56b47b1 fix: handle _workspace placeholder in section routing (greptile review) (ElasticClaw Bot)
7fea1af feat: review and improve static-export settings route coverage (ElasticClaw Bot)
13eb009 Address provider status review feedback (Marc Campbell)
9bac3cb fix: make factory claws default to PR completion (Ana Berg)
970047a Remove local provider (Marc Campbell)
1e10de3 Merge pull request #352 from elasticclaw/feat/docker-web-env (Ana Berg)
93b3204 fix: address Docker dev review findings (Ana Berg)
b9868bd fix: replace Ollama catalog patch script with Go (Ana Berg)
8eebce0 chore: restore Next route type reference (Ana Berg)
32b2852 fix: stabilize Docker agent Ollama bootstrap (Ana Berg)
238e461 feat: support Ollama in hub AI settings (Ana Berg)
c7b2c52 feat: add optional Ollama dev stack (Ana Berg)
3eeac76 feat: add Docker-based local dev environment (Ana Berg)
bf79bef feat: register Docker provider checks (Ana Berg)
72ef5cd feat: wire Docker provider into hub provisioning (Ana Berg)
54a1c56 feat: add Docker sandbox provider (Ana Berg)
738b001 Address LLM retry review feedback (Marc Campbell)
7e3d399 Avoid resending delivered LLM turns (Marc Campbell)
619cae8 Retry transient LLM upstream errors (Marc Campbell)
b1df737 Merge pull request #344 from elasticclaw/fix-context-overflow-recovery (Marc Campbell)
120c860 Merge pull request #347 from elasticclaw/fix-github-issue-card-title (Marc Campbell)
4ee9517 Merge pull request #349 from elasticclaw/feat/atomic-external-webhook (Marc Campbell)
457ecff Scope external trigger claim cleanup (Marc Campbell)
b1a9805 Merge pull request #348 from elasticclaw/fix-e2e-tracker-isolation (Marc Campbell)
0e7f4f7 Retry ngrok reserved domain creation (Marc Campbell)
108fd1e Address E2E isolation review feedback (Marc Campbell)
a8f688e Address issue title review feedback (Marc Campbell)
f3e49d1 Respect caller cancellation during session recovery (Marc Campbell)
55dacd6 fix: make external webhook trigger creation atomic (Claw)
ffed8d3 Isolate concurrent E2E tracker runs (Marc Campbell)
7da437e Merge pull request #345 from elasticclaw/feat/workflow-harness-tests (Marc Campbell)
29ac6fe Link GitHub issue agent titles (Marc Campbell)
4027313 fix: address greptile review feedback on PR #345 (ElasticClaw Agent)
5fc20f4 fix: store claw token on FakeBridge for checkpoint completion (ElasticClaw Agent)
a5308ac Recover bridge sessions after context overflow (Marc Campbell)
a73bf85 feat: add workflow harness integration tests (ElasticClaw Agent)
03f18c8 fix: add db migration comment for pipeline_outputs table (Claw)
1d56434 feat: workspace scripts with workflow output capture (#342) (Claw)
6b8d992 Merge pull request #341 from elasticclaw/fix-stuck-review-injections (Marc Campbell)
7f0b35e Merge pull request #339 from elasticclaw/feat/daytona-resilience-331 (Marc Campbell)
8d5fa49 Merge pull request #340 from elasticclaw/enforce-initial-agent-plan (Marc Campbell)
bf83a59 Drain queued injection after send failure (Marc Campbell)
5c6abcd Include initial plan gate in pipeline injects (Marc Campbell)
b08fa99 Address injected message review feedback (Marc Campbell)
0a3ce79 Address Daytona resilience review feedback (Marc Campbell)
d9d68d7 Address initial plan review feedback (Marc Campbell)
9746e32 Queue injected messages behind active turns (Marc Campbell)
318f240 convince the agent to send the initial plan (Marc Campbell)
2d13504 feat: make Daytona workflow runs resilient to workspace, tool-policy, and model timeout failures (Claw)
5812c0d fix: remove no-op new-session reset action from web UI (Claw)
4d1b8df Merge pull request #337 from elasticclaw/workflow-enable-controls (Marc Campbell)
eab314a Address workflow controls review feedback (Marc Campbell)
ba72cce add pause buttons to workflows (Marc Campbell)
5c2352e Merge pull request #335 from elasticclaw/fix/cross-tab-auth (Marc Campbell)
76d2377 Address cross-tab auth review feedback (Marc Campbell)
1059c32 Require visible workflow progress updates (Marc Campbell)
35c31ab fix: use localStorage for auth tokens to enable cross-tab sessions (ElasticClaw Agent)
5937380 Address activity payload and Daytona bootstrap issues (Marc Campbell)
9e424b4 Address activity transparency review feedback (Marc Campbell)
b823f86 Improve agent activity transparency (Marc Campbell)
2368579 fix: make web lint script self-contained (Claw)
6bc845d Address missing label test review (Marc Campbell)
f346a32 Ignore missing GitHub labels on removal (Marc Campbell)
d719faf Address manual trigger review feedback (Marc Campbell)
a5bc827 add manual trigger to workflow, small test (Marc Campbell)
c5ea9f1 Address Greptile trigger validation comment (Marc Campbell)
432075a update workflow validation to support new trigger shape (Marc Campbell)
7cee2b8 Avoid known host first-use race (Marc Campbell)
440b6f1 Stabilize Linear E2E workflow triggers (Marc Campbell)
ef2ddb8 Honor replicated workflow run timeouts (Marc Campbell)
255eb44 Trust SSH host keys on first use (Marc Campbell)
c79c47f Disable replicated workflow run actions (Marc Campbell)
6d59cd6 Reuse server SSH helper for workflow runs (Marc Campbell)
127a2e9 Add workflow run action (Marc Campbell)
5b458d4 Address Replicated E2E review feedback (Marc Campbell)
afdd28d Add Replicated CMX E2E coverage (Marc Campbell)
251127c Document E2E config as secrets (Marc Campbell)
a1c8de4 Retry Daytona E2E cleanup conflicts (Marc Campbell)
fa21857 Harden Linear E2E webhooks (Marc Campbell)
1f888b8 Add Linear Daytona E2E suite (Marc Campbell)
003d12f Add E2E Daytona finalizer (Marc Campbell)
91ce818 Harden E2E sandbox cleanup (Marc Campbell)
f1e0fea Filter and isolate E2E runs (Marc Campbell)
09a7760 Add initial PR E2E harness (#320) (Marc Campbell)