Skip to content

Family 5: slow-burn temporal — composite-identity actor continuity across sessions #45

Description

@NickCrew

Thesis

The slow-burn / temporal scenario. The same attacker returns over a long
window
, each session individually unremarkable; the attack is only visible when
you stitch the sessions across time via composite identity (JA4 + JA4H + auth
token — the "actor continuity even when cookies are cleared" property Synapse's
own site claims). Recon on Monday, a probe Wednesday, the real attempt Friday.

This is the hardest family to build (needs real time or time-simulation) and
the most damning if Synapse misses it, because persistent cross-session actor
continuity is an explicit product claim.

What it probes in Synapse

Does the actor model persist and accumulate risk across sessions, or reset
when the session/cookie resets? Does composite identity (JA4 + JA4H + token)
re-link a returning actor who cleared cookies?

Build challenge — time

The scenario needs either a long real wall-clock window or a time-simulation
hook so Monday/Wednesday/Friday can be compressed. Open question to resolve in
this issue: does Synapse expose a clock/time-injection seam for testing, or must
the harness (atlas-crew/Bridge#3) drive real elapsed time? Flag this before
building — it determines feasibility.

Honesty constraints

  • Each session is individually unremarkable — no single session is blockable.
  • The actor is re-linkable only by composite fingerprint, not by cookie (clear
    cookies between sessions to actually test the claim).
  • No metadata announces "same actor" — Synapse must re-identify by behavior+fp.

Ancestors / adjacent

directory-enumeration-slow.json, vp-demo-03-behavioral-trigger.json,
apt-multi-vector-campaign.json.

Acceptance criteria

  • ≥3 individually-unremarkable sessions separated in time, cookies cleared
    between them.
  • Actor re-linkable only via composite identity (JA4 + JA4H + token).
  • Time mechanism resolved (real window vs. simulated clock) and documented.
  • Reported as cross-session risk accumulation: did Friday's attempt inherit
    Monday/Wednesday's risk, or start cold?

Related

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    Projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions