Poisoned Pipeline

Poisoned Pipeline / Home

Poisoned Pipeline

A catalogue of software supply-chain incidents: 333 documented cases, 117 techniques, and the ecosystems they hit.

333 incidents
117 techniques
93 observed in incidents
24 identified from research

Incidents per year

1975198319941999200220032007200820092010201120122013201520162017201820192020202120222023202420252026

Hover a bar for the count. 2025–2026 is the worm/ATO wave (Shai-Hulud and descendants); the pre-2020 bars are the classic vendor-updater era.

Most-used techniques

T0109166
T0184110
T017283
T012170
T015958
T012945
T015144
T013636
T014036
T011134
T011331
T020929

By number of incidents mapping to each technique. Full list in the technique matrix.

Eras

54 pre-2020 classics (CCleaner, NotPetya, ShadowHammer era)
48 2020–2024 (extension and registry hijack era)
231 2025–2026 worm/ATO wave

The 2025–2026 wave is roughly one incident every three days per StepSecurity's tracking.

Frequently seen in write-ups

Actor / malware names appearing most often across incident summaries.

Browse

Technique matrix — techniques × tactics, heat-mapped by incident count

CI/CD matrix — Mercari's pipeline matrix, with the incidents that actually do that

Incident mapping — every incident → techniques

Platforms — all incidents for npm, PyPI, GitHub Actions, …

Attack stories — chronological list

Project

Guidance mapping — OWASP / NIST / SLSA / CISA

Origins — where each technique came from

Conference talks — supply-chain and AI talks, mapped or gapped

Sources — the watch list this catalogue tracks

Changelog — what changed, and when