Pillar 1 v1.1 (shipped earlier) flagged benign browsing bursts as
beaconing: on short captures, grid periods longer than the observed
span never complete a cycle, so significance saturates to 1 and the
score sticks at exactly 0.5. v1.2 caps the grid at a third of the
observed span, requires a minimum number of completed cycles, and
excludes protocols that are periodic by nature (NTP, DNS, mDNS,
ICMP). Validated against 300 seeds per scenario: false positives on
bursty traffic drop from the 290-300/300 that v1.1 produced to 0/300,
with every beacon-detection scenario scoring unchanged.
Add Pillar 2 (persistent/long-lived sessions, with Zeek's fragmented
TCP sessions rejoined after a 5-minute gap), Pillar 3 (DNS tunneling
via Shannon-entropy information volume per subdomain per hour; the
Jensen-Shannon divergence was measured and does not separate tunnels
from benign CDN subdomains, so it stays in the evidence at zero
weight instead of in the score), and Pillar 4 (deterministic
noisy-OR fusion across the three pillars).
The fusion formula was validated against 8 labeled real-traffic
vectors (IoT-23, CTU-Normal) with tolerance <= 1e-4. A known
limitation is kept, not hidden: a benign Philips HUE bulb's strict
periodic polling and a 21.9h session score it above three real C2
hosts. No weight or threshold was tuned to mask this; it is
documented in the architecture docs instead.
Also fixes real bugs found along the way: ingest_dir's partition
count summed file names instead of distinct dates, DuckDB errors
were counted as bad rows, /api/alerts always returned an empty list
regardless of the last analysis, and the dashboard still claimed
every pillar was unimplemented.
Restores four design corrections that were missing from
docs/architecture.md (KL to Jensen-Shannon, the persistent-session
detector, the JA4 licensing boundary, and deterministic reduction
order), and catalogs the public datasets used for validation
(IoT-23, CTU-Normal, Mendeley DTQBC) with their licenses and SHA-256
checksums under research/validation/.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VAyr8d7HuAdUGB6aA5xmDD
Acauã: Network Detection & Threat Hunting, heuristic analysis over a
rolling 7-day window using deterministic math (time series, graph
theory, information theory) instead of opaque ML models. Third
product in the Ciberfauna ecosystem, alongside Guará and Sabiá.
Follows the ecosystem's established conventions: OpenWolf project
memory, bilingual PT-BR/EN docs and UI, Gitea Actions CI/CD on the
same self-hosted runners, Conventional Commits, SemVer, logo and
productName handling, MIT OR Apache-2.0 licensing.
Architecture decision (made with the project owner after a dedicated
consultation, not the original blueprint's suggested stack): Rust
throughout, not Python/FastAPI/React. Most of the statistics (Bowley
skew, MADM, Shannon entropy, PCR, Z-score, novel-edge detection,
pivot chains) reduces to SQL over an embedded DuckDB/Parquet store;
only the spectral analysis and graph centrality need host-language
code, and both have workable Rust crates. Documented in full, including
six corrections to the original blueprint's math (Rayleigh/Schuster
instead of a degenerate Lomb-Scargle over bare timestamps, an explicit
sentinel for Bowley skew's Q3=Q1 case, Jensen-Shannon instead of
unbounded KL, a dedicated long-connection detector, JA4 vs. JA4+
licensing caution, deterministic per-alert evidence), in
docs/architecture.md.
Skeleton only: crates/acaua-core has the four detection pillars,
fusion, storage, zeek log parsing and the updater as typed stubs that
compile and return a clear "skeleton" status rather than real
detection logic — that is separate, future work. crates/acaua-cli has
serve/analyze/doctor/update. apps/gui is a minimal Tauri shell. Also
already includes the one-click desktop self-update mechanism shared
with Guará and Sabiá (AppImage/.deb/Windows installer), even with no
release published yet, for consistency across the ecosystem.
No Gitea repository exists for this project yet; this is a local-only
commit.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VAyr8d7HuAdUGB6aA5xmDD
