Ulasan Lengkap Mengenai Jenis Pertaruhan Menggunakan E Wallet Peranti Salah

Dark GitHub-style web UI built with Next.js 16, Tailwind 4, shadcn/ui components, Motion and dnd-kit: dashboard, live sessions and transcript viewer, run launcher with live headless logs and tmux console, agents, kanban board with drag and drop, templates, project pages and settings, plus a Ctrl+K command palette. Adds a Playwright smoke test that boots the production build against a fake ~/.claude and fake claude binary and checks auth, DNS rebinding, CSRF, CSP, XSS, drag and drop, launching runs and mobile layout. Adds README, SECURITY.md and the final kanban. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

- zod 4 keeps .default() values inside .partial(), so every PATCH (including kanban drag & drop) reset description, labels, checklist, prompt, status and priority. Use a dedicated update schema without defaults. - runIds is server-managed: no longer writable via PATCH. - POST /api/runs validates taskId before spawning a process. - Run log follows output by scrolling its container, not the window. - Regression tests: unit, API and E2E smoke. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

Replace the GitHub palette with neutral near-black surfaces, subtle borders, a light primary CTA and a monochrome logo/nav indicator. All hard-coded GitHub colours in components are now semantic tokens (link, orange, closed, border-strong) defined once in globals.css. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

Runs typecheck, lint, unit/integration tests, production build, the Playwright E2E smoke test (screenshots uploaded as artifact) and a production dependency audit on every PR and push to main. Read-only token, no persisted credentials, cancelled on superseded pushes. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

LayoutProps/PageProps are global types produced by Next's typegen; on a fresh checkout (CI) tsc ran before they existed and failed. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

…(v0.2.0) - Capture a git baseline (HEAD, pre-existing changes) before each run without touching the repo, and expose GET /api/runs/:id/diff with commits, numstat, new untracked files and a size-capped patch; new 'Cambios' tab with a diff viewer. - Route every git call through a hardened helper that neutralises filters, textconv, external diff, fsmonitor, hooks and pagers configured by the repository (including .git/info/attributes); git status on project pages could previously run a malicious clean filter. - Toast + optional desktop notification when a run finishes (also for runs shorter than the poll interval); toggle in settings. - Release workflow: when CI passes on main, tag vX.Y.Z from package.json and publish a GitHub Release with notes from CHANGELOG.md. Version shown in UI. - Tests: run diff, hostile repo, E2E diff tab, notification and version. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

Large or unreadable untracked sets (e.g. an unignored node_modules) overflowed git's output buffer, leaving an empty baseline so every pre-existing file looked created by the run. Use a larger buffer, and above 5000 entries mark the set as unknown instead of storing it in runs.json; the diff then only reports tracked changes and says so. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

Aggregate token usage from local Claude Code transcripts per UTC day, project, model and tool, with per-file caching by (mtime, size) and bounded parsing concurrency. New /analytics page: 7/30/90-day ranges, project filter, stat tiles, a single-series daily column chart with hover breakdown and a table view, plus ranked bars by project, model and tool. Cost only shows the real figures reported by Agentic runs; no price estimates. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

A long session that started before the window inflated the model and tool breakdowns with usage from outside it. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

- TaskQueue runs To do tasks with a project one at a time by priority, moves successful ones to review, and stops on limit, empty queue or first failure. Hard per-batch limit, optional per-task budget, never bypassPermissions, never resumes after a restart. API: /api/queue, /start, /stop; tasks page gets a preview dialog and a live progress banner. - SessionIndex.search scans prompts and Claude replies newest-first with a raw-line prefilter, hit limit and time budget; sessions page gets an 'En el contenido' mode with highlighted snippets. - Tests: queue ordering/limits/failures/restart, search, E2E for both. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

The client sent POSTs without a body and without a JSON content type, which the server's CSRF check rejects, so 'Detener' and 'Parar tras la actual' never worked from the UI. Mutations now always send a JSON body; E2E test clicks 'Detener' on a live run. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

- Local credential store for Slack, GitHub, Bitbucket and Jira (~/.agentic/integrations.json, 0600); the API only ever returns masked hints and keeps stored secrets when a field is left empty. - Hardened outbound client: https only, no credentials in URLs, requests pinned to the configured origin, same-origin redirects only (max 3, never for writes), 10s timeout, 5MB cap, readable 502s. - /integrations page to connect, test and disconnect each provider. - Slack via bot token: channels, history, threads, send and reply from the new /messages page; Agentic notifications for finished runs and queue (bot channel or incoming webhook, optional failures-only) with mrkdwn escaping of user text. - Tests against a local Slack stand-in; E2E connects Slack from the UI and replies in a thread. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

- PullService normalises GitHub and Bitbucket PRs: unified inbox from the origin remotes of known projects, detail with description, comments, reviews/participants, checks/statuses and per-file patches. - Actions: comment, approve, request changes, merge (squash/merge/rebase or fast-forward) pinned to the reviewed head SHA, close/decline. Only repos of known projects; owner/repo/number validated. - UI: /pulls inbox with filters and /pulls/view detail with conversation, files (diff viewer) and checks tabs, confirm dialogs, external links only if http(s), and 'Revisar con Claude' launching a review run. - Tests against GitHub/Bitbucket stand-ins (incl. same-origin redirect for Bitbucket diffs); E2E approves and merges a PR from the app. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

409 (head moved) and 405 (required checks, reviews or branch protection) used to surface as 'GitHub: error 409'; now they explain what to do. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

…ude (v0.7.0) - JiraService on the Jira Cloud REST v3 API: JQL search via the current /search/jql endpoint, issue detail with ADF description and comments rendered as plain text, comments posted as ADF, transitions verified against the ones Jira currently offers, idempotent import as a linked task with mapped priority. - /jira page: JQL bar with presets and configurable default, issue list and side panel to comment, change status, import or 'Resolver con Claude'. - Tests against a Jira stand-in; E2E transitions, comments and imports an issue from the app. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

- RunManager.streamLog tails the NDJSON log as SSE: one event per line, byte offset as event id so EventSource reconnects resume via Last-Event-ID without duplicates, 15s heartbeat, 'end' event when the run is over, aborts with the request. - GET /api/runs/:id/stream (same auth/host checks as the rest of the API). - Run page consumes the stream with per-frame batching, shows 'en vivo', and falls back to polling if SSE is unavailable. - Tests for ordering, resume and abort; E2E asserts lines arrive while the run is still running in the production build. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

- Server-side sessions: 256-bit random ids, only SHA-256 stored, 30-day TTL and 7-day idle expiry; the master token only opens a session (or Bearer API access) and the cookie never contains it. - Passkeys (WebAuthn via @simplewebauthn): platform authenticator with required user verification (Touch ID), single-use 5-minute challenges in an HttpOnly cookie, counters updated, RP 'localhost'. - 'Exigir Touch ID' disables token URL and Bearer access and closes token sessions; only enabled from a passkey session; 'npm run reset-auth' for local recovery. - /login page outside the app shell (route group), logout, session list and revocation, rate limiting on token and passkey login. - Audit log (audit.log, 0600, rotated): logins, failures, passkey and session changes, every successful API mutation (method + path only). - Tighten ~/.agentic permissions at startup; access URL uses localhost. - Tests for sessions, expiry, revocation, require-passkey rules, challenges, rate limit, audit and permissions; E2E registers Touch ID with Chromium's virtual authenticator, logs out, signs in with it and enforces it. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

- Electron shell around the exact same UI: runs the Next.js standalone server with Electron's embedded Node on 127.0.0.1, loads it from localhost; hardened window (context isolation, sandbox, no Node in the renderer, external links to the default browser, notifications only). - Native Touch ID unlock that opens a 'desktop' server-side session (kept when Touch ID is required); inherits the login shell PATH so claude, git and tmux are found when launched from the Dock. - In-app updates without an Apple Developer ID: checked on every entry (launch, activate, focus; throttled to 10 min), from the app menu and from Settings via a minimal contextBridge; downloads the zip for the CPU from the latest GitHub Release, verifies SHA-512 from latest-mac.yml and the bundle signature, swaps the bundle with rollback and relaunches; private repos use the GitHub token from Integrations. - electron-builder config (dmg + zip, arm64 + x64, ad-hoc or Developer ID signing, optional notarization); Desktop workflow builds and launches the app on macOS; releases attach dmg, zip and latest-mac.yml. - Tests for desktop sessions, env helpers and updater logic. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

…29.0) - "/" menu in the Headless reply box and the Chat: /plan, /edits, /ask, /auto, /model handled by Agentic (mode/model for the message), plus the user's and project's custom commands and skills from ~/.claude and <project>/.claude, sent to Claude Code as typed. Chat gains a permission-mode selector (never bypassPermissions). - Library: indexes artifact, PR, issue and document links found in session transcripts (streamed, cached by mtime/size, newest first, time budget), aggregated by URL with context, project and date. Co-Authored-By: Claude Sonnet 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

…eshacer) Co-Authored-By: Claude Sonnet 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

…tas por estado Co-Authored-By: Claude Sonnet 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

… de ejecución ni en el chat Co-Authored-By: Claude Sonnet 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

… de Headless Co-Authored-By: Claude Sonnet 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

… por proyecto Co-Authored-By: Claude Sonnet 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

…puestas en Markdown Co-Authored-By: Claude Sonnet 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

Co-Authored-By: Claude Sonnet 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

…o adaptable Co-Authored-By: Claude Sonnet 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

…ivo real de los errores de Bitbucket/GitHub Co-Authored-By: Claude Sonnet 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

…el chat Co-Authored-By: Claude Sonnet 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

Co-Authored-By: Claude Sonnet 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

…o y Touch ID sin saltar solo Co-Authored-By: Claude Sonnet 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

…tado en la mitad derecha Co-Authored-By: Claude Sonnet 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP

… que ocurren, con su código Co-Authored-By: Claude Sonnet 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01NsWdNH7fzHCnFbRsNYoBwP