Threat actor
GhostAction
Last fetched
GhostAction is a threat actor tracked in WhisperGraph's MITRE ATT&CK corpus, observed using 0 techniques.
Description
Software supply chain campaign and its unnamed operator, first reported by GitGuardian in September 2025: malicious GitHub Actions workflows named "Github Actions Security" injected into compromised repositories exfiltrated CI/CD secrets, with 3,325 secrets stolen from 817 repositories. In October 2026 OpenSourceMalware reported a renewed wave and a second stage that no longer needs a CI runner: a .vscode/tasks.json launcher with a folderOpen trigger runs a Node.js credential stealer disguised as public/fonts/fa-solid-400.woff2, installs a cron job polling the operator's server, and is backed by a fallback workflow. OpenSourceMalware calls this stage a copycat of the DPRK PolinRider technique rather than an extension of it, and ties it to GhostAction through shared server infrastructure. A phishing kit on the lookalike domain my-github.com, harvesting credentials and silently authorizing an OAuth application with repo and workflow scopes, was reported on the same infrastructure.
Techniques by tactic
No ATT&CK techniques are recorded for GhostAction in WhisperGraph.
Attributed infrastructure
None published. WhisperGraph carries no ATTRIBUTED_TO edge to GhostAction today — this states the absence of a published link, not that GhostAction has no infrastructure.
References
© The MITRE Corporation. This work is reproduced and distributed with the permission of The MITRE Corporation.
Related pages
Pivot from GhostAction into its techniques, tactics and any attributed infrastructure.
Queries
Resolves the slug to this actor, merging every duplicate node sharing the same name.
MATCH (a:ACTOR)
WHERE a.name =~ $pattern OR any(x IN a.aliases WHERE x =~ $pattern)
RETURN a.id AS id, a.name AS name, a.aliases AS aliases, a.description AS description,
a.references AS references, a.campaigns AS campaigns
LIMIT 25Run yourself →Techniques this actor uses, grouped by the tactic each one serves.
MATCH (a:ACTOR {name: $name})-[:USES_TECHNIQUE]->(t:ATTACK_PATTERN)
OPTIONAL MATCH (t)-[:USES_TACTIC]->(tac:ATTACK_PATTERN)
RETURN t.id AS techniqueId, t.name AS techniqueName, tac.id AS tacticId, tac.name AS tacticName
LIMIT 1000Run yourself →Infrastructure publicly attributed to this actor.
MATCH (n)-[:ATTRIBUTED_TO]->(a:ACTOR {name: $name})
RETURN labels(n)[0] AS kind, n.name AS name
LIMIT 25Run yourself →Or query Whisper from your own LLM workflow via the Whisper MCP server.