/chain

Supply-Chain Malware Handoffs

Purpose: Canonical handoff payloads for Chain's live supply-chain malware recipes. Read when: A lockfile, IDE hook, runtime, git-history, CI, or maintainer-publish finding must move between Chain and a domain owner.

All payloads follow _common/HANDOFF.md. Emit one target per handoff, reference concrete evidence, defang attacker URLs, and never include credential values.

Inbound

USER_TO_CHAIN_MALWARE_REQUEST

from: User
to: Chain
intent: malware-scan | campaign-scan | lockfile | eradicate | rotate | harden | propagation
context:
  os: macOS | Linux | Windows | WSL | container | CI
  suspect_window: "<UTC range>"
  suspect_campaign: "<campaign or unknown>"
  scope: ["<repo path, host scope, image, or runner>"]
  prior_actions: "<actions already taken>"
constraints:
  read_only_until_confirm: true
  may_quarantine: false

SENTINEL_TO_CHAIN_MALWARE_REQUEST

from: Sentinel
to: Chain
trigger: malicious_pin | slopsquat_candidate | install_script_anomaly
context:
  package: "<name@version>"
  lockfile_path: "<path>"
  advisory: "<CVE, GHSA, or source URL>"
  confidence: HIGH | MEDIUM | LOW
ask:
  - Confirm live-environment persistence, droplet, and process evidence
  - Return infection grade and evidence chain

BUILDER_OR_TRAIL_TO_CHAIN_MALWARE_REQUEST

from: Builder | Trail
to: Chain
trigger: lockfile_diff | optional_dependency | install_script | suspicious_commit | force_pushed_tag
context:
  ref_or_commit: "<sha>"
  changed_files: ["<path>"]
  author: "<if relevant>"
ask:
  - Cross-check exact evidence against the campaign IoC database
  - Block merge only on an evidence-backed malicious match

TRIAGE_TO_CHAIN_MALWARE_REQUEST

from: Triage
to: Chain
trigger: supply_chain_incident
context:
  incident_id: "<id>"
  severity: SEV1 | SEV2
  affected_targets: ["<host, runner, repo, or image>"]
constraints:
  approved_actions: ["<pre-authorized persistence-stop action>"]
  blocked_actions: ["credential revocation", "production infrastructure changes"]
ask:
  - Run a full IoC scan and classify the infection
  - Stop IoC-matched persistence if actively bleeding and authorized
  - Return evidence chain, eradication gate, and rotation eligibility

Outbound

CHAIN_TO_TRIAGE_MALWARE_INCIDENT

from: Chain
to: Triage
trigger: CONFIRMED | ACTIVELY_BLEEDING
context:
  campaign: "<name>"
  grade: CONFIRMED | ACTIVELY_BLEEDING
  evidence:
    persistence: [{path: "<path>", sha256: "<hash>", status: stopped | running}]
    lockfile_pins: [{package: "<name@version>", path: "<lockfile>"}]
    passive_exfil_traces: [{host: "<defanged host>", evidence: "<log line>"}]
  quarantine_path: "/tmp/chain-malware-quarantine-<utc>/"
  eradication_status: in_progress | verified_clean | blocked
  rotation_status: not_eligible | ready | issued
  retaliation_risk: "<why rotation remains gated>"

CHAIN_TO_SENTINEL_MALICIOUS_PIN

from: Chain
to: Sentinel
trigger: confirmed_malicious_version_pin
context:
  package: "<name@version>"
  lockfiles: ["<path>"]
  clean_version: "<name@version>"
  campaign: "<name>"
ask:
  - Find the pin across the organization
  - Propose an upgrade and rollback plan
  - Add a slopsquat rule if applicable

CHAIN_TO_GEAR_MALWARE_HARDENING

from: Chain
to: Gear
trigger: verified_clean_rebuild
context:
  affected_surfaces: [npm_runtime, github_actions, container_image, dependency_bot]
  controls:
    - "ignore-scripts=true"
    - "minimum release age of at least 7 days"
    - "full-SHA GitHub Action pinning"
    - "OIDC instead of long-lived tokens"
    - "registry proxy and provenance checks"
ask:
  - Rebuild runners and images from a clean base
  - Apply controls and return verification evidence

CHAIN_TO_VIGIL_MALWARE_RULES

from: Chain
to: Vigil
trigger: new_source_cited_ioc
context:
  campaign: "<name>"
  signatures:
    - type: filename | sha256 | process_cmdline | network_endpoint | git_pattern
      value: "<exact or defanged value>"
      source: "<advisory URL + date>"
ask:
  - Author and test Sigma or YARA coverage
  - Map the evidence to current MITRE ATT&CK techniques

CHAIN_TO_LORE_MALWARE_PATTERN

from: Chain
to: Lore
trigger: repeated_campaign_pattern | novel_persistence | eradication_order_lesson
context:
  pattern_summary: "<short, non-sensitive summary>"
  recurrence_count: <n>
  affected_agents: [Sentinel, Chain, Vigil, Triage, Gear]
ask:
  - Curate the reusable pattern without raw credentials or confidential payloads

Handoff Hygiene

• One target per handoff; do not combine Triage, Sentinel, Gear, and Vigil requests.
• Include exact paths, hashes, timestamps, package pins, and source citations.
• Defang attacker URLs (https to hxxps, .com to [.]com) and use passive evidence only.
• Preserve persistence stopped before revoke and verify-clean before rotation as explicit gates.
• A confirmed IDE-hook compromise remains within Chain: quarantine only after approval, then rerun intake and regenerate .chain-manifest.json on the cleaned directory.

Overlap Boundaries