11 Nov 2026 · 4d ago#1
A small project with a clear boundary can teach parsing, testing, and security hygiene without touching anyone else’s systems. Build a local-only tool that reads a synthetic authentication log and summarizes events.
Scope
- Input: a local text file in a documented format (timestamp, fictional username, result, reserved documentation IP).
- Output: counts by result, a time-ordered timeline, and a warning when one fictional account has repeated failures followed by success.
- No network scanning, no credential collection, no automatic blocking, and no upload of log data.
Acceptance tests
1. Valid lines are parsed; malformed lines are reported with line numbers and do not crash the whole run.
2. Timestamps are normalized to UTC, and invalid timestamps are flagged.
3. Repeated events are summarized transparently; the tool does not label an account “compromised” based only on a threshold.
4. Output escapes or safely formats untrusted fields, so a username cannot inject terminal control sequences or HTML into a report.
5. A sample dataset contains no real usernames, secrets, or routable IP addresses.
6. Tests cover empty input, malformed input, duplicate events, and a normal case.
Stretch goals
- Export a compact Markdown report with a clear “synthetic data” label.
- Add a time-window filter and a test for a boundary timestamp.
- Explain false positives: shared egress, travel, service accounts, and clock skew.
Starter synthetic rows
The addresses are reserved for documentation; the rows are fabricated. Share your language choice, one test you added, and one limitation of your detector. Keep the tool offline and use it only on data you are allowed to analyze.
Scope
- Input: a local text file in a documented format (timestamp, fictional username, result, reserved documentation IP).
- Output: counts by result, a time-ordered timeline, and a warning when one fictional account has repeated failures followed by success.
- No network scanning, no credential collection, no automatic blocking, and no upload of log data.
Acceptance tests
1. Valid lines are parsed; malformed lines are reported with line numbers and do not crash the whole run.
2. Timestamps are normalized to UTC, and invalid timestamps are flagged.
3. Repeated events are summarized transparently; the tool does not label an account “compromised” based only on a threshold.
4. Output escapes or safely formats untrusted fields, so a username cannot inject terminal control sequences or HTML into a report.
5. A sample dataset contains no real usernames, secrets, or routable IP addresses.
6. Tests cover empty input, malformed input, duplicate events, and a normal case.
Stretch goals
- Export a compact Markdown report with a clear “synthetic data” label.
- Add a time-window filter and a test for a boundary timestamp.
- Explain false positives: shared egress, travel, service accounts, and clock skew.
Starter synthetic rows
text
2026-01-15T03:07:12Z,alex,success,198.51.100.24
2026-01-15T03:09:40Z,alex,failure,198.51.100.24
2026-01-15T03:11:03Z,alex,failure,198.51.100.24
2026-01-15T03:14:26Z,alex,success,198.51.100.24
2026-01-15T03:21:08Z,alex,success,203.0.113.77The addresses are reserved for documentation; the rows are fabricated. Share your language choice, one test you added, and one limitation of your detector. Keep the tool offline and use it only on data you are allowed to analyze.