Skip to content
KitploitKITPLOIT
ToolsExploitsBlog
Log in
Submit
ToolsExploitsBlog
Submit

Hacking, PenTest, and Cybersecurity Tools for Your Security Arsenal!

Kitploit is a directory of hacking, cybersecurity, and pentesting tools. Discover the latest project updates to find vulnerabilities, analyze systems, automate testing, and strengthen your security.

FeedsContactPrivacy© 2026 Kitploit

Tool Directory

Categories

View all categories
Loading categories
Tools/GitHubGitHub/afsharid/gitea-cve-2026-116289-detection
Authentication & AuthorizationDefensive ToolsVulnerability AnalysisIdentity & Access Management (IAM)Papers & ResearchIncident ResponseLog AnalysisLabs & Practice
Share
GitHubafsharid/gitea-cve-2026-116289-detection

gitea-cve-2026-116289-detection

Sigma detection rules, lab and findings for CVE-2026-116289 (Gitea public-only PAT to org Actions runner)

View Repository
2 days agoNot yet reviewed

Detection rules for CVE-2026-116289 (Gitea public-only PAT → org Actions runner)

Sigma rules for CVE-2026-116289 / GHSA-qcgm-ggp8-22qc, built and tested in a local lab against Gitea 28.2.0 (vulnerable) and 28.3.0 (patched).

SeverityCVSS 3.1 6.5 Medium (AV:N/AC:L/PR:H/UI:N/S:U/C:H/I:H/A:N), per the official advisory
AffectedGitea ≤ 28.2.0
Fixed28.3.0 (commit 01db50f0bf)
WeaknessCWE-863 Incorrect Authorization

The vulnerability in one paragraph

A personal access token scoped public-only,write:organization that belongs to an org owner could request an org-level runner registration token for a public organization. The runner registered with it did not inherit the public-only restriction, so it could receive jobs, secrets and a GITEA_TOKEN from the org's private repositories. The fix rejects public-only tokens in reqOrgOwnership() with HTTP 403.

Rules

All rules are in rules/combined.yml (logsource: product: gitea, service: router).

RuleLevelWhat it shows
Registration token issued via APIlowHunting/audit. On ≤ 28.2.0 it fires on abuse and on legitimate provisioning.
Registration token request rejected (403, reqOrgOwnership)mediumOn ≥ 28.3.0: a public-only token or a non-owner tried to mint an org runner token.
Runner Register RPCinformationalBase rule and runner inventory.
Token issued → runner registered on same instance (correlation)lowTime-only pairing on one instance.

Findings

Lab-verified on 2026-11-18 unless marked otherwise.

  1. Reproduced. The same POST /api/v1/orgs/{org}/actions/runners/registration-token with a public-only,write:organization PAT returns 200 with a token on 28.2.0 and 403 this endpoint is not available for public-only tokens on 28.3.0. A normal write:organization PAT returns 200 on both.
  2. Indistinguishable on vulnerable versions. In the router log the public-only call and the normal call have the same method, path, status and handler. Only the source port and latency differ.
  3. The fix is visible in the handler field. On 28.3.0 the rejected call is logged as @ v1/api.go:500(v1.reqOrgOwnership.1). The public-only check runs inside reqOrgOwnership, so the log names that middleware for both public-only tokens and non-owners.
  4. Handler line numbers drift between versions (tokenRequiresScopes is api.go:329 on 28.2.0 and :335 on 28.3.0). Rules match function names, never file:line.
  5. Runner registration is logged. POST /api/actions/runner.v1.RunnerService/Register appears with status 200 on both versions.
  6. Token minting and runner registration come from different source IPs in the normal flow (lab: host 192.168.97.1 vs runner container 192.168.97.4). A correlation that groups by src_ip never fires.
  7. Path variants still match. A query string (?x=1), a trailing slash and a different org-name case (PublicOrg) all return 200 and are logged with the raw path.
  8. Direct private-repo access with the lab PAT returned 403, not the 404 in the advisory. The lab token lacked read:repository, so the scope check fired first.
  9. Pre-upgrade artefacts are not revoked (inferred from the advisory and the fix diff, not lab-tested). The fix rejects new requests only. Runners and registration tokens created before the upgrade remain valid.
  10. Severity. The official advisory scores 6.5 Medium. Some third-party digests list this CVE as 9.1 Critical; use the advisory. The advisory text says "Gitea 1.27.0" while its affected range is <= 28.2.0.

What the logs can and cannot prove

  • The Gitea router log does not record token scopes. On vulnerable versions, the public-only call and a normal call produce identical lines (lab-verified).
  • Token minting and runner registration come from different source IPs even in the normal flow (lab: host vs runner container). The correlation therefore groups by instance, not by src_ip.
  • Private-job delivery is only provable from the database (access_token, action_runner, action_task joined with repository.is_private).
  • After patching, runners and registration tokens created before the upgrade are not revoked by the fix. Audit org-level runners and reset org registration tokens.

Lab and tests

cd lab && docker compose -p giteacve up -d   # 28.2.0 on :3082, 28.3.0 on :3083
./lab/repro.sh                                # registration-token step only: 200 vs 403
tests/check.sh                                # rules vs logs/router-sample.log
uv tool run --python 3.12 --from sigma-cli sigma check rules/combined.yml

repro.sh only requests a registration token. It does not register a runner or take any job or secret. lab/seed.env holds lab-only credentials and is not committed.

Test results (2026-11-18): sigma check 0 errors / 0 issues; rule hits 7 / 1 / 2; correlation 5 pairs. Path variants with a query string, a trailing slash and a different org-name case also match. tests/run_rules.py is a minimal evaluator for the modifiers these rules use. It is not a SIEM backend; test the rules in your own SIEM.

Not lab-tested: the GET, repo, user and admin registration-token routes; the admin-route branch of the 403 rule; behaviour behind a reverse proxy.

AI assistance

The analysis and rule drafts were produced with Claude Fable 5.2 through the Claude API (research/), then tested in the lab and corrected. The lab tests found two issues in the drafts: the src_ip grouping of the correlation and an outdated ATT&CK tactic tag.

License

Rules and documentation: Detection Rule License (DRL) 1.1.

Download Tool

Most Popular

View all →

Discover the most used tools by our community.

  • 1

    OsintHunter

    letox4/osinthunter

    🎯 The ultimate OSINT toolkit: investigation graph, confidence scoring, alias families, map, timeline and reports. 49 modules, 400+ sites, no API keys required.

    1.1k
  • 2

    ghostlock-app

    yukonga/ghostlock-app

    GhostLock One-Tap Execution App (CVE-2026-43499)

    603
  • 3

    looking-glass

    anonghosty/looking-glass

    CSIRT platform that ingests Shadowserver reports into MongoDB and replays them as interactive maps, analysis, findings, advisories and executive dashboards.

    390
  • 4

    ARTEX

    baarai/artex

    LLM multi-agent autonomous penetration testing platform with planner/worker agents, asset and exploration graphs, MITM traffic recording, and vulnerability retesting.

    321
  • 5

    CyberStrikeAI

    ed1s0nz/cyberstrikeai

    The system of action for AI-native cybersecurity—where intent becomes governed execution, evidence becomes operational memory, and every operation improves the next.

    301
  • 6

    redforge

    cybersecurity91/redforge

    Browser-side red team toolkit with 102 offline tools for recon, payload generation, credential attacks, crypto, encoding and engagement reporting.

    263

Explore all tools

Browse our collection of tools

View all tools →