CVE Vulnerability Catalog

atomic-agents-stack before 1.1.0 contains a cost-guardrail bypass in the _estimate_batch_cost function that returns zero cost for unknown models not in the pricing table. Attackers can configure deployments with unknown model identifiers to bypass daily cost caps and exceed budget limits in parallel batch operations.

gitoxide gix-transport before 0.59.2 fails to filter control characters in git-daemon connect requests, allowing attackers to inject NUL/CR/LF bytes via crafted git URLs. Attackers can inject extra NUL-delimited protocol fields to spoof virtual hosts or inject newlines into daemon requests and logs.

Vikunja before 2.6.0 fails to validate that user-supplied project_view_id in task-position requests belongs to the task's project. Authenticated attackers can insert task position rows into arbitrary other tenant project views via POST or PUT task-position endpoints.

Vikunja before 2.6.0 contains an API token scope bypass vulnerability in task read endpoints where authorization fails to inspect query string parameters. Attackers with limited token scopes can use the expand parameter to access restricted data like comments, reactions, and time entries without proper permission verification.

Vikunja before 2.6.0 continues to expose the raw TOTP shared secret after enrollment through the GET /api/v1/user/settings/totp and /api/v1/user/settings/totp/qrcode endpoints without re-authentication. Attackers with a valid access token can read the secret, import it into their own authenticator, and generate valid codes indefinitely to defeat the second factor and enable account takeover.

Vikunja versions before 2.6.0 fail to properly validate link-share tokens in the v2 API user search endpoints. Attackers with a read-only share link can enumerate project users via the projects endpoint and confirm arbitrary usernames exist via the global search endpoint.

vikunja before 2.6.0 fails to validate team access when attaching teams to projects, allowing authenticated users to enumerate all teams and members. Attackers can attach arbitrary team IDs via the project teams endpoint to retrieve complete team rosters including member names and admin flags for unauthorized teams.

Vikunja before 2.6.0 fails to limit archive expansion during data import, allowing authenticated users to cause denial of service. Attackers can upload highly compressed files that expand to tens of gigabytes in memory and disk, exhausting server resources and crashing the instance.

Vikunja before 2.6.0 fails to apply pixel decode limits to avatar and project-background upload endpoints, allowing authenticated users to upload crafted images that decode to excessive pixel counts. Attackers can upload small images with extreme aspect ratios that consume significant CPU and memory during processing, causing denial of service through repeated or concurrent uploads.

Vikunja versions before 2.6.0 contain a resource exhaustion vulnerability in the Planka migrator that fails to enforce aggregate memory budgets during migration jobs. Authenticated attackers can submit migration requests pointing to attacker-controlled servers advertising numerous size-compliant attachments, exhausting worker memory and causing denial of service for all users.

vikunja versions before 2.6.0 contain a resource exhaustion vulnerability in the POST /api/v2/migration/csv/migrate endpoint that fails to limit parsed row cardinality. Authenticated attackers can upload multipart CSV files with millions of tiny records to exhaust process memory and terminate the API service.

vikunja versions before 2.6.0 contain a resource exhaustion vulnerability in the task-filter endpoint that accepts deeply nested filter expressions without recursion depth limits. Authenticated attackers can supply thousands of nested parentheses in the filter query parameter to exhaust memory and terminate the API process.

AVideo through 29.0 contains a blind server-side request forgery vulnerability in the getHeaderContentTypeFromURL function that issues get_headers() calls guarded only by format validation. Authenticated users with canUpload permission can store attacker-chosen URLs as video links, triggering vulnerable function execution on every video watch page render to probe internal hosts.

AVideo through 29.0 contains an unauthenticated server-side request forgery vulnerability in the check_site_availability function that accepts attacker-controlled HTTP Host headers. Attackers can send requests to submitIndex.php or ajax.php with arbitrary Host headers to probe internal network hosts and ports, following redirects without authentication.

FreeRDP versions before 3.31.0 contain an uninitialized heap memory disclosure vulnerability in the urbdrc USB redirection channel. A malicious RDP server can induce failing USB transfers to read uninitialized heap memory from the client, defeating ASLR and enabling remote code execution when chained with memory corruption vulnerabilities.

FreeRDP before 3.31.0 contains an integer overflow in the audin Apple backends when processing FramesPerPacket values from MSG_SNDIN_OPEN messages. Attackers can supply crafted FramesPerPacket values that cause AudioQueueAllocateBuffer size computation to wrap, resulting in undersized buffer allocation and potential out-of-bounds access.

FreeRDP before 3.31.0 contains a denial-of-service vulnerability in the URBDRC control-transfer request path that fails to validate OutputBufferSize before forwarding to the libusb backend. A malicious RDP server can send a control-transfer request with OutputBufferSize set to 65536, triggering a reachable assertion that terminates the client process.

FreeRDP versions before 3.31.0 contain an integer overflow in WinPR's Stream_EnsureRemainingCapacity function that allows remote attackers to cause denial of service. A malicious RD Gateway peer can send a WebSocket Ping frame with a crafted 64-bit extended payload length to trigger integer wraparound, resulting in a double free that crashes the FreeRDP client during connection.

FreeRDP before 3.31.0 contains a buffer over-read vulnerability in the rts_read_result function within the RPC gateway transport parser. Attackers can send a malicious BIND_ACK PDU with a truncated result entry to trigger an out-of-bounds read causing process abort.

FreeRDP versions before 3.31.0 fail to validate MonitorIds array values when parsing RDP connection files, allowing unbounded array indexing in xf_detect_monitors. Attackers can craft a malicious RDP file with an out-of-range selectedmonitors value to trigger out-of-bounds heap read and write operations when opened in xfreerdp.