awslabs/aidlc-workflows v2.11.1-preview.20261016.1

AI-DLC Workflow 2.11.1-preview.20261016.1

on GitHub

Warning: Full Suite failed for this source. A Full Suite failure report ends these notes.

Warning: Updating from the last release to this preview failed in its checks. Stay on your release or wait for the next preview.

Merged commits since preview source 9ecefa0:

• fix(re-entry): a bare /aidlc in a new chat asks again the question you were asked and never answered (#2726)
• test(live drives): a golden journey on opencode to Reverse Engineering approved, after one change the person asks for (#2725)
• fix(sensor): a sensor's script stays in the tools folder on Windows too (#2695) (#2697)
• test: isolate t242 gate words fixture (#2703)
• fix: update smol-toml to 1.9.1 (#2696)
• chore(deps-dev): bump @xterm/headless from 5.5.0 to 6.0.0 (#2677)
• chore(deps): bump the github-actions-minor-and-patch group with 2 updates (#2679)
• chore(deps): bump actions/setup-node from 6.5.0 to 7.1.0 (#2681)
• chore(deps-dev): bump zensical (#2678)
• test(live drives): a golden bugfix journey on Copilot CLI, with the plan question left overnight and approved in a new session (#2719)
• fix(engine): next names the step that finishes an open review, from one function, in every state the review log leaves it in (#2587) (#2688)
• fix(codekb): one spelling of a scope path in publish and compare, so a trailing slash is not a refusal (#2586)
• fix(config): a section's conflict names the config command that takes --force (#2718)
• fix(plugins): plugin list and doctor name engine plugin sync for a plugin that needs composing (#2716)
• fix(answers): a verification command question asked in one chat is answered in a new one, in one reply (#2715)
• fix(plugins): turning a plugin off through config project takes out what it added, as plugin select does, and no longer stops on a file the person never touched (#2713)
• fix(stop-hook): after the review is recorded the agent goes on to the approval gate instead of ending its turn (#2400) (#2588)
• fix(review): the summary confirmation names only the files it will generate, not the questions file the person just answered in (#2711)
• test(live drives): a golden bugfix journey on Codex, with the plan question left overnight and approved in a new session (#2710)
• fix(plugins): a team plugin composes after the one base it builds on; a plugin that cannot is held with one line while the rest compose, and turning a base off pauses the plugin built on it (#2693) (#2709)
• fix(review): the summary confirmation no longer says why it is asked now (#2708)
• fix(stages): a copy on 2.8.0 to 2.11.0 refreshes to the next release instead of stopping with "unknown key: cites" (#2707)
• fix(kiro-ide): a Unit's reviewer reads only that Unit, as a preview feature (#2637)
• test(live drives): a person's pick on a lone Claude Code menu goes into the drive's record in its own words (#2692)
• fix(guards): a stage reported without its outputs or code, a source changed after its review, or the review cap part way through a Unit's step is the agent's own step, not a question for the person (#2685)
• fix(status): on a clone with work here and none selected, status names each piece and where it stands (#2690)
• fix(type-check): on a native install without Bun, your project's own TypeScript checks your types instead of every check passing unchecked (#2686)
• fix(engine): no "check the rebuild-stage-graph hook" line when the Units were read right from their dependency file (#2672)
• fix(engine): words beside finished work name it and the way back to a stage, instead of calling them new work (#2669)
• fix: refresh Python dependency graph (#2656)
• fix(engine): in a chat that has not joined the work, a refused report names the steps to join it and goes on, instead of a dead end (#2670)
• fix(switches): turning a check back on from the chat never changes the team's committed settings (#2665)
• fix(opencode): two chats in one window each act on their own work (#2667)
• ci: enable weekly Dependabot updates on main (#2674)
• fix(review): side by side, a Unit's reviewer writes its review where the verdict reads it, so the batch can land (#2666)

Source commit: d27f96b

Update from the last release

Updating from the last release to this preview failed in the preview run.

Nightly test report

Full Suite failed for d27f96b6ab26 in run 38968296482. The preview build does not wait for these tests.

Failed legs: native_terminal (failure), native_reconcile (failure), live_linux (failure).

Failing tests

• t-exec-codex-compose-front.serial (1 failed assertions) in live-codex-2-Linux
  • t-exec-codex-compose-front - interactive compose over exec + exec resume > beat 1 stops at the gate with nothing written; beat 2 resume-approves and creates the intent
• t-live-aidlc-answers-corpus.serial (1 failed assertions) in live-claude-sdk-1-Linux
  • claude: the answers corpus through a real agent > rs-02 resume-reentry chat ""

Runner errors

• native-darwin-bun: INCOMPLETE: required selected coverage was not fully exercised
• native-darwin-bun: ERROR: test matrix job did not fulfill its planned coverage; see test-matrix-receipt.json
• native-linux-bun: INCOMPLETE: required selected coverage was not fully exercised
• native-linux-bun: ERROR: test matrix job did not fulfill its planned coverage; see test-matrix-receipt.json
• native-windows-bun: INCOMPLETE: required selected coverage was not fully exercised
• native-windows-bun: ERROR: test matrix job did not fulfill its planned coverage; see test-matrix-receipt.json

Failed jobs

• Linux / claude-sdk (1): Linux / claude-sdk 1/5
• Linux / codex (1): Linux / codex 2/2
• Linux / native-reconcile (1): Linux / native-reconcile
• Linux / native-terminal (1): Linux / native-terminal bun
• Linux / result (1): Linux / result
• macOS / native-terminal (1): macOS / native-terminal bun
• Windows / native-terminal (1): Windows / native-terminal bun

Full logs are in the run 38968296482 full-suite-* artifacts.