Bound Linux builds and qualify live installations
- Keep Linux dashboard builds within their memory budget.
- Install approved changes for live qualification while preserving unfinished cycle checks.
- Keep release verification read-only and check prepared edge files against the approved source.
Full detail
- **** Restrict stylesheet discovery to authored sources, use the normal webpack compiler worker, and preserve the four-GiB build limit with shared-host headroom and exact workload diagnostics.
- **** Qualify an approved installation while its cycle remains open, preserving immutable partial reports and pending live criteria. Release proof startup keeps workspace bindings read-only and verifies independently prepared edge files against the exact source.
- ** (partial):** Keep representative and cached-release timing criteria open until measured against real installations.
Check deployment proof before preparation
- Explain release blockers before preparing an installation.
- Keep dashboard builds and workspace tests within their resource budgets.
- Record deployment phases without counting an already-live check as a new release.
Full detail
- **** Run one read-only publication proof before installation preparation, bind acceptance to immutable build heads, and collect independent blocking checks before mutation.
- **** Preserve Next's normal compiler worker, require sufficient shared-host headroom, and contain dashboard compilation within its workload budget.
- **** Route supported workspace tests and watch commands through the canonical runner with package discovery and at most four workers.
- ** (partial):** Record install phases and classify already-live repeats separately. Representative and cached-release merge-to-live timing remain unproven.
Keep accepted releases moving
- Preserve accepted evidence when reviewed changes are integrated and merged.
- Include the Cycle 90 review, release and large-board improvements.
Full detail
- **** Accept literal catch-all route paths and independently reviewed same-PR producer integration in release proofs, preserving original receipts and rejecting missing or reverted work. Includes the Cycle 90 changes from v0.402.108, whose activation was blocked before tagging.
Keep reviews, releases and large boards moving
- Accept qualified changes and attach installation evidence to the same report later.
- Verify approved fixes across merged pull requests and their reviewed integration.
- Browse large boards with accurate counts, saved views and bounded pages.
Full detail
- **** Accept qualified self-hosted changes before activation and attach later production verification to the original build report without rebuilding.
- **** Verify each approved squash landing against the explicit remote release target, preserving original build and review evidence.
- **** Select board pages, ordering, counts, groups and facets in one authorized PostgreSQL query through both adapters.
- **** Use bounded database pages in dashboard board views, preserve saved views and Realtime refresh, and load off-page task details separately.
- **** Bind Windows test memory accounting and scoped cleanup to process creation time so reused IDs cannot claim unrelated applications.
- **** Recognize explicitly retired edge workers within the existing bounded safe-read recovery, preserving generic-error and write safeguards.
- **** Isolate bulk browser-fixture preparation from Realtime subscriptions and verify the exact project channel before testing change delivery.
Keep the dashboard aligned with your latest changes
- Show the final realtime change without a manual refresh.
- Save rapid preference changes in order and make failures visible.
- Keep your latest project selection and clear deleted projects.
Full detail
- **** Deliver the last realtime refresh in a burst and permit the configured API's matching WebSocket origin.
- **** Serialize preference writes, retain the latest intent, bind writes to the signed-in user and show persistence failures.
- **** Ignore stale project responses, retain valid selection on read failure and clear selection after deleting the final project.
Keep task changes and their evidence together
- Retry build completion without losing evidence, and keep every review round.
- Undo resets safely and assign tasks through consistent ownership safeguards.
- Keep build reports in the task's own cycle and avoid concurrent report ID collisions.
- Return rejected builds to In Progress for rework.
Full detail
- **** Save a build report, its task transition and audit events in one transaction; retain a stable identity for safe retries.
- **** Preserve feedback from later review rounds and return the actual committed task status on replay.
- **** Snapshot and cancel the same locked tasks, then undo unchanged rows and consume the snapshot atomically.
- **** Use the canonical ownership action for hub assignments without changing task status or cycle.
- **** Record the task's exact cycle UUID and member separately from its builder; require explicit cycle membership for dashboard completion.
- **** Allocate concurrent report IDs inside the completion transaction through the existing numeric database allocator.
- **** Return rejected builds to In Progress while recording the verdict and transition together.
Retry failed payment updates safely
- Retry payment updates after a database failure without losing the account change.
- Apply concurrent payment retries once.
Full detail
- **** Commit payment receipts, account entitlements and conversion history together so database failures remain retryable and concurrent retries apply once.
Keep concurrent planning sessions separate
- Apply the preparation from your own planning session, even when teammates plan at the same time.
- Retry a failed apply without losing its prepared context.
- Assign selected tasks to your cycle when generating handoffs separately.
Full detail
- **** Bind hosted plan, handoff, strategy and cycle-append preparations to the authenticated caller and a server-issued token. Keep concurrent flows separate, expire abandoned state, and preserve failed applies for retry. Hosted plans that skip handoffs now assign every selected task to the matching member cycle.
Preserve shipped evidence and follow consistent workflow guidance
- Restore missing task evidence while preserving its original owner and cycle.
- Keep customized skills visible in each selected AI client.
- Use consistent planning, build and review guidance across supported source bundles.
Full detail
- **** Restore missing evidence for already-shipped tasks through a scoped, append-only record that preserves the original assignment, cycle and effort.
- **** Validate skill installation destinations before writing, preserve local overrides and expose the installed native skills.
- **** Generate planning, build, review and strategy guidance from one authored source across supported distribution paths.
Find workflow guidance and classify decisions deliberately
- Installed AI clients get a concise pointer to the same PAPI workflow guide.
- Owners can review legacy decision types in small batches and confirm each choice.
- Unclassified decisions stay visible, with counts and pages selected in the database.
Full detail
- **** Setup and skill installation add one canonical how-to-work-with-PAPI pointer to each selected harness's eager instructions, preserving local rules.
- **** Owners can queue legacy decision classification in batches of ten or confirm a type on a decision card. Database-scoped counts, ordering, pagination and transactional audit entries keep unclassified records visible and prevent silent changes.
Plan together and reconcile releases reliably
- See collaborators, pointers and decision history on the same cycle brief.
- Browser and AI clients join the exact shared plan without mixing teammate cycles.
- Approved identical squash merges reconcile without changing historical build reports.
Full detail
- Cycle 86 — Shared Planning and Faster Delivery
- Share the exact cycle brief with live collaborators and bounded pointers. Browser and MCP participants use the same canonical cycle ID, with existing Active Decision history showing authors and times.
- Accept an approved squash landing when its complete source tree matches the reviewed build, including a final documentation-only commit. Preserve the original redacted restore-receipt error. Historical build reports stay unchanged; the ten-minute full and repeat release measurements remain outstanding.
Shared project settings and clearer setup
- Team settings brings module ownership, conventions, briefs and decision editing together.
- Setup explains each step in plain or technical language and marks starting decisions as provisional.
- Equivalent release checks can finish without cancelling each other.
Full detail
- Cycle 85 — Shared Context and Faster Delivery
- Consolidate module ownership, conventions, brief and decision editing in Team settings. Project members can edit shared context; member administration, billing and security retain their existing permissions. Decision edits preserve history and reject concurrent stale saves.
- Explain setup artifacts in sequence, offer plain or technical explanations, and identify starting decisions as provisional.
- Prevent equivalent push and merged-PR CI runs from cancelling each other. The ten-minute merge-to-live target still requires representative and cached-repeat deployment measurements.
Safer project actions and simpler release verification
- Dashboard actions consistently use the project you are viewing.
- Release verification recognizes approved squash merges without changing historical build reports.
- Search has a measured baseline for relevance and response times.
Full detail
- Cycle 84 — Faster Reliable Project Operations
- Bind remaining dashboard comments, votes, feedback, milestones, recommendations, learning resolution and connection reads to the viewed authorized project while preserving existing permissions.
- Verify reviewed task inclusion in an approved squash merge without rewriting historical build reports. The ten-minute merge-to-live target remains subject to measured deployment evidence.
- Measure document and history retrieval relevance, latency and default query plans through PostgreSQL and the real hosted transport using a bounded synthetic corpus.
Reliable releases and consistent cycle data
- Release recovery handles temporary dispatch failures and reuses verified artifacts.
- Cycle membership updates preserve concurrent changes across the dashboard and MCP.
- Read and manage your own planning-log entries directly through MCP.
Full detail
- Cycle 83 — Reliable Releases and Cycle Data
- Discover landed work from fresh remote main, recover bounded transient workflow dispatch failures, and reuse verified immutable artifacts without rebuilding. Real ten-minute release timing remains to be measured.
- Delegate cycle membership from the dashboard, PostgreSQL adapter and hosted service to one transactional RPC, preserving concurrent roster edits.
- Read, edit and delete planning-log entries through MCP using exact entry IDs and author-scoped mutations.
Safe build recovery and less repeated work
- Completion failures tell you whether a report was saved before suggesting recovery.
- Planning keeps useful capability evidence without repeating identical descriptions.
- Release builds use stable cache paths and avoid duplicating runtime dependencies.
Full detail
- Cycle 82 — Faster Reliable Delivery
- Distinguish report writes that were not attempted, succeeded, or have an uncertain outcome before suggesting recovery.
- Collapse repeated capability descriptions while retaining distinct file information, changed deliverables, and task references.
- Use stable locked build staging, avoid a duplicate dependency-tree copy, bound streamed inventory hashing to four files, and yield while waiting for the exact qualified artifact. The ten-minute target still requires measured release evidence.
More reliable upgrades, with clear failure and waiting records
- Upgrades use the approved installer so installer fixes apply during the upgrade itself.
- Task lists and releases read less data while retaining task and cycle checks.
- Runner waits remain separate from failed installations, and original failure details are preserved.
- Release timing stops at verified live health, before later bookkeeping.
Full detail
- Upgrade through the approved candidate installer, preserving installer fixes during activation; artifact qualification now exercises the real activation preflight.
- Bound release integrity reads to the selected cycle and its Done tasks; board pages fetch heavy detail only for the displayed tasks.
- Treat queued CI as waiting, preserve original activation errors, and upload failed attempt receipts beside the immutable artifact.
- Measure verified live time before reconciliation, record admin health correctly, and avoid unrelated branch scans for an explicit release target.
Install skills for the AI clients you use
- Choose native skill directories explicitly while keeping local overrides and edited skills intact.
- Review context includes project conventions and relevant decisions for the reviewer to consider.
- Build attribution works through the existing MCP connection and keeps unknown identity valid.
- Validated release stages remain intact until activation.
Full detail
- Cycle 81 — Client compatibility and review context
- Install the pinned skill bundle into explicitly selected native directories for supported AI clients, preserving local overrides and edits.
- Deliver plugin-free attribution guidance on connect and require stronger completion evidence to name its exact turn or message.
- Include project conventions and relevant decisions in review context as findings for the reviewer to consider.
- Keep validated release stages intact until activation.
Failed installs are visible, and releases reach the live check sooner
- A failed install leaves one compact receipt with its source, elapsed time, stopping phase and redacted error.
- The read-only migration ledger check runs before artifact qualification, catching missing migrations before the expensive build.
- Detailed build diagnostics remain in the workflow log, while deployment receipts focus on health, configuration and rollback.
- Build reports keep runtime model evidence separate from caller self-reports.
Full detail
- Cycle 79 — Release reliability and attribution
- Failed installs now leave one compact receipt with the source commit, elapsed time, failed phase and redacted error; build details stay in the linked workflow logs.
- Check the live migration ledger before artifact qualification, so an unapplied migration fails before the slow build and before activation.
- Keep the ten-minute merge-to-live target and retain only the activation evidence needed to verify configuration, health checks and rollback.
- Build reports distinguish exact runtime model evidence from a caller's self-reported model.