Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 2 additions & 3 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -84,9 +84,8 @@ flood can delay or hide a real attacker that appears only once. The limit resets
An alert fires when an IP's flag enters the alert set (`--alert-on`, default
`veneer,paradox,threat`) or changes, not on every sighting. Alerts go to stdout (one JSON object
per line with `--json`) and, with `--webhook`, are POSTed as JSON. The payload's `text` field works
as-is with Slack and Mattermost incoming webhooks. Log lines are untrusted, so
the attacker-written `sample` line is kept out of `text` and only appears in the JSON `alert`
object. `docker stop` and Ctrl+C shut it down cleanly. It exits 1 if anything alerted.
as-is with Slack and Mattermost incoming webhooks. Log lines are untrusted and can hold private
details, so the `sample` line appears only in `--json` stdout, never in `text` or a webhook post. `docker stop` and Ctrl+C shut it down cleanly. It exits 1 if anything alerted.

Developing: `python3 -m unittest discover -s tests` (no keys or network needed).

Expand Down
8 changes: 5 additions & 3 deletions ROADMAP.md
Original file line number Diff line number Diff line change
Expand Up @@ -407,13 +407,15 @@ VirusTotal's 500 for rescans.
on restart. Not measured against a real log, so 15 may be too low for a busy one; the watcher warns on stderr when it
hits the limit, once per stretch at the limit.

### 4.7 Keep raw log lines out of webhook posts · S
### 4.7 Keep raw log lines out of webhook posts · S · ✅ shipped

- [ ] **PDX-34: Webhooks don't send log lines.** Done when: a test shows a webhook POST contains no `sample` and none of the log line's text, stdout's `--json` output is unchanged, and the README and `SECURITY.md` say so.
- [x] **PDX-34: Webhooks don't send log lines.** Done when: a test shows a webhook POST contains no `sample` and none of the log line's text, stdout's `--json` output is unchanged, and the README and `SECURITY.md` say so.

The JSON `alert` includes the raw log line, which can hold usernames, internal hostnames or tokens, and it goes to a
third party.

*Downside:* a receiver that read `alert.sample` no longer gets it. `--json` stdout still has it for local use.

### 4.8 Webhook URL out of the command line · S

- [ ] **PDX-35: Read the webhook URL from the environment.** Done when: `PARADOX_WEBHOOK` sets the URL, a test shows it's used without `--webhook`, and the README's examples use it and say why.
Expand Down Expand Up @@ -549,7 +551,7 @@ and fix the top issues. Repeat after 5.6. It also doubles as class material.
| When | Items | Why first |
|---|---|---|
| Now | ~~1.1, 1.2, 1.5, 3.2, 5.1~~ (done) | Small, keyless, and fix the biggest measured weakness (the scanner rule's dependence on GreyNoise); 5.1 makes that dependence visible |
| Next | 1.3, 3.1, 3.5, 3.6, 4.1, 4.7, 4.8, 4.9, 4.10, ~~4.6, 4.5, 1.7, 1.8, 3.7, 1.4, 1.6, 5.2, 5.4~~ (done) | Close the known false positives, build adversarial and hiding-place test sets, test the AI summarizer under attack, and show watch state and claim checks in the UI |
| Next | 1.3, 3.1, 3.5, 3.6, 4.1, 4.8, 4.9, 4.10, ~~4.6, 4.7, 4.5, 1.7, 1.8, 3.7, 1.4, 1.6, 5.2, 5.4~~ (done) | Close the known false positives, build adversarial and hiding-place test sets, test the AI summarizer under attack, and show watch state and claim checks in the UI |
| Then | 2.1, 2.3, 3.3, 3.4, 4.2, ~~4.3, 4.4~~ (done), 5.3, 5.5–5.7 | New unmasking signals plus self-measurement once history has accumulated, with the UI restructured to hold them |
| Later | 2.2, 2.4, 2.5 | Bigger extensions: domains, campaigns, more beacons |

Expand Down
8 changes: 3 additions & 5 deletions SECURITY.md
Original file line number Diff line number Diff line change
Expand Up @@ -100,15 +100,13 @@ more than people expect.

**What's in place:**

- The Slack-compatible `text` field holds only the flag, IP, scores and verdict. The attacker-written `sample` log line
is kept out of it.
- The Slack-compatible `text` field holds only the flag, IP, scores and verdict. The raw `sample` log line is removed
from everything posted, and appears only in `--json` stdout on your own machine.
- Webhooks are opt-in, and a failed post is reported on stderr without stopping the watcher.

**Open:**

- The JSON `alert` object does include `sample`, the raw log line (up to 300 characters), along with hit counts. A log
line can hold usernames, internal hostnames or tokens, and any receiver that stores the whole payload gets it. Slack
and Mattermost use `text`, but they still receive the rest.
- A post still says which IPs were seen, how often, and the log's path (`source`), and the receiver can store all of it.
- `text` can quote a third-party label in the verdict, so a chat channel or bot reading it sees attacker-influenced text.
- The URL is passed on the command line, so it's visible in process listings and shell history. It's effectively a
password for that channel.
Expand Down
4 changes: 3 additions & 1 deletion paradox/watch.py
Original file line number Diff line number Diff line change
Expand Up @@ -228,7 +228,9 @@ def alert(self, row: dict, previous: str | None) -> None:
with self.lock:
print(line, file=self.out, flush=True)
if self.webhook:
threading.Thread(target=self._post, args=(payload,), daemon=True).start()
# A log line can hold usernames, hostnames or tokens, and this goes to a third party.
outbound = {**payload, "alert": {k: v for k, v in row.items() if k != "sample"}}
threading.Thread(target=self._post, args=(outbound,), daemon=True).start()

def _post(self, payload: dict) -> None:
req = urllib.request.Request(self.webhook, data=json.dumps(payload).encode(), method="POST",
Expand Down
20 changes: 20 additions & 0 deletions tests/test_app.py
Original file line number Diff line number Diff line change
Expand Up @@ -109,6 +109,26 @@ def test_scans_once_and_alerts_on_flagged_ips_only(self):
self.assertIn("PARADOX 45.1.1.1", alerts[0]["text"])
self.assertEqual(w.seen["45.1.1.1"]["hits"], 3)

def test_webhook_gets_no_log_line_but_stdout_keeps_it(self):
w, _, out = self.watcher({"45.1.1.1": fake_score(paradox=50, threat=0.5)}, webhook="https://example.test/x")
w.feed("Failed password for token-s3cret from 45.1.1.1", "auth.log")
posted, sent = [], threading.Event()

def urlopen(req, timeout=None):
posted.append(req.data.decode())
sent.set()
return mock.MagicMock()

with mock.patch("urllib.request.urlopen", urlopen):
w.check(w.q.get())
self.assertTrue(sent.wait(5))
self.assertNotIn("s3cret", posted[0])
body = json.loads(posted[0])
self.assertNotIn("sample", body["alert"])
self.assertEqual(body["alert"]["ip"], "45.1.1.1")
self.assertIn("PARADOX 45.1.1.1", body["text"])
self.assertIn("s3cret", json.loads(out.getvalue())["alert"]["sample"])

def test_new_ip_cap_stops_a_flood_and_recovers(self):
w, _, _ = self.watcher({}, max_new_per_hour=3)
with mock.patch("paradox.watch.time.time", return_value=1_800_000_000.0) as clock, \
Expand Down
Loading