v0.8.15: ci improvements, trigger machine resizing, connectors fixes

[waleedlatif1]

• improvement(api): retire route contracts that no route serves (improvement(api): retire route contracts that no route serves #7206)
• improvement(ci): bound docker layer caches and right-size ten runners (improvement(ci): bound docker layer caches and right-size ten runners #7210)
• improvement(background): right-size the two knowledge task machines (improvement(background): right-size the two knowledge task machines #7212)
• improvement(ci): scan CodeQL PRs at the promotion boundary, refresh main daily (improvement(ci): scan CodeQL PRs at the promotion boundary, refresh main daily #7213)
• improvement(ci): bound the local Turborepo cache (improvement(ci): bound the local Turborepo cache #7214)
• fix(connectors): honor provider retry deadlines (fix(connectors): honor provider retry deadlines #7211)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
docs	Skipped		Aug 28, 2026 9:41am

[greptile-apps]

Greptile Summary

This release improves CI cache bounds and runner sizing, resizes two knowledge background tasks, narrows and refreshes CodeQL scanning, removes contracts for retired routes, hardens Atlassian discovery validation, and propagates provider retry deadlines into connector scheduling.

• Adds bounded Docker and Turborepo cache configuration and right-sizes CI runners.
• Replaces retired HTTP route contracts with direct in-process schemas and adds a route-contract audit command.
• Consolidates retryable HTTP error construction and carries provider retry timing into connector failure scheduling.
• Tightens Atlassian accessible-resource payload validation.

Confidence Score: 4/5

The PR appears safe to merge, with a non-blocking cleanup needed in the new route-audit script's logging and error handling.

The changed runtime paths preserve internal schema validation and bounded connector backoff behavior; the only accepted issue is that the new developer audit command bypasses repository-standard diagnostics utilities.

Files Needing Attention: scripts/check-api-contract-routes.ts

Important Files Changed

Filename	Overview
apps/sim/lib/knowledge/documents/utils.ts	Centralizes bounded retryable HTTP errors, response cleanup, and retrieval of validated retry delays through error cause chains.
apps/sim/lib/knowledge/connectors/sync-engine.ts	Propagates provider retry deadlines into connector scheduling without weakening the existing failure backoff or one-day ceiling.
apps/sim/lib/internal/tool-operations/parse-contract-input.ts	Generalizes internal request parsing so retired HTTP operations can validate against bare schemas while preserving contract callers.
scripts/check-api-contract-routes.ts	Adds a route/contract consistency audit, but its diagnostics bypass required shared logging and error-normalization utilities.
.github/workflows/ci.yml	Bounds Docker caches and adjusts runner sizes across development, release, and desktop jobs.
.github/workflows/codeql.yml	Moves PR scanning to the main promotion boundary, refreshes main daily, and sizes runners by analyzed language.
apps/sim/lib/atlassian/discovery.ts	Validates accessible-resource response shape and distinguishes malformed responses from credentials with no site grants.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart LR
  Provider[Connector provider] --> Fetch[Retrying fetch layer]
  Fetch --> Error[Retryable HTTP error]
  Error --> Delay[Validated retryAfterMs]
  Delay --> Sync[Connector sync failure handler]
  Sync --> Backoff[Max of failure ladder and provider delay]
  Backoff --> Schedule[nextSyncAt, capped at one day]

Loading

_{Reviews (1): Last reviewed commit: "fix(connectors): honor provider retry de..." | Re-trigger Greptile}