Savepoint Arc
Official Savepoint Arc website

Intrusion detection help for your own network on Windows

Savepoint Arc 2.9.30 is an AI agent for Windows 10 and 11. It reads your IDS alert exports, edits rule files with a diff you approve, and runs your own test commands.

Version

2.9.30

Released 8 Nov 2026

Setup

34 MB

The download button gives this file

Portable

25 MB

Single executable, separate download

Languages

13

Interface languages of the app

Models

3 kinds

Cloud, local or built-in

System

Windows

Windows 10 and 11, x64

Alert review

Intrusion detection work an agent can take over

Intrusion detection means watching traffic or host activity and raising alerts. Alerts need a human decision. Savepoint Arc supports that decision work on your own network: it reads exported alerts, sorts them into expected activity, misconfiguration or something to investigate, and proposes rule changes in a file you review. The setup program is about 34 MB.

01

Sort a day of alerts into three groups

Your sensor exported yesterday's alerts to a folder on your PC, and you want a short list of what needs attention. list_dir shows the export files, read_file opens the newest one, and grep pulls out every line marked high. Nothing else runs. The agent then writes a triage note with write_file, one line per alert, and the card lists the saved file. Network intrusion detection like this works on exports, because the agent does not watch live traffic or block anything. Switch on Ask first before the write, read the 3 lines it proposes and change any verdict you disagree with.

Tool trace
list_dir  exports/  4 files
read_file  exports/alerts-latest.csv  18 rows
grep  high  3 matches
write_file  notes/triage.md  saved
02

Tune host based intrusion detection system rules

One rule keeps firing on a machine you administer, and the log shows the activity is expected. Ask the agent to read the rule file with read_file and the last 5 alerts with grep, then to propose a change with edit_file. The card shows a diff, so you see the exact lines before anything is saved. Nothing is saved yet. Afterwards Shell runs your own configuration test and reports the real exit code on the card. The same steps suit intrusion detection and prevention systems, because their rules are plain text files too.

Tool trace
read_file  rules/local.rules  read
grep  rule-id  5 matches
edit_file  rules/local.rules  +2 -1
Shell  your config test  exit 0
03

Review network based intrusion detection system exports

A sensor on your own network segment writes alerts to a log, and you want to know which rule is the noisiest. Start with exports. grep filters one rule name at a time, read_file shows the lines around a match, and write_file saves a summary that lists each rule with your decision beside it. Sensors can produce thousands of lines, so work in slices of one rule or one hour. Before you accept the summary, open it, compare 2 or 3 entries with the raw log and correct what looks wrong.

04

Write down triage steps for the next shift

Alert triage means deciding for each alert whether it is expected activity, a misconfiguration or something to investigate, and writing down why. Ask the agent to create a notes folder with make_dir and a decision log with write_file. Later, grep finds earlier decisions for the same rule. Lookup is easy. Start with Plan only when a rule set is new to you: the agent writes the steps as a plan and changes nothing, and you switch to Ask first when you want it to save the 4 notes.

Read the workflow guide

How it differs from a typical chat assistant

It acts. Version 2.9.30 runs your commands and edits your files itself, then shows every real result as a card in the chat.

Typical chat assistantSavepoint Arc
CommandsDescribes a command for you to runRuns it in Shell and reports the exit code
FilesPastes edited text into the chatEdits the file and shows a diff
ApprovalNothing to approveAuto-run, Ask first or Plan only
EvidenceText you must verify yourselfA card per action with its real result

How to get started

  1. Download the setup program (about 34 MB) and compare its SHA-256 checksum with the published manifest.
  2. Run it on Windows 10 or 11, then open the app. No account is needed.
  3. Connect a model: a cloud provider, a local model or the built-in model, then press Test.
  4. Choose the folder with your exported alerts and rule files, pick Ask first and describe the task.

Who uses it and when

Beginners learning IDS basics start here. Go slowly. Network monitoring for small offices is a common first job, with one administrator, one sensor and an export folder that nobody has had time to read, so the agent summarises it while you check each line.

Practitioners who tune rules every week use it for the repetitive part. If you are building an IDS project using machine learning, keep the labels in a CSV file and let the agent compare them with the exported alerts. Rules stay yours.

Administrators of a few hosts use it to document decisions. They run Plan only first, read the plan, then switch to Ask first so that each of the 3 or 4 edits waits for approval. Reasons get written down. Adaptive IDS rules can work in practice only when a person reviews every change.

Questions people ask first

What is intrusion detection?

Intrusion detection is the practice of watching network traffic or host activity and raising alerts about suspicious behaviour. An IDS only alerts. The decisions stay with you, and Savepoint Arc helps by reading exported alerts, grouping them and recording the reason for each decision so a noisy rule can be tuned later.

Is syriancode.com the official Savepoint Arc website?

Yes, syriancode.com is the official Savepoint Arc website. Compare the SHA-256 checksum of the installer you download with the published manifest, using the certutil command on Windows, before you run it. Check the hash. Questions go through the Send feedback button inside the app. This site records the time, IP address and browser of each download click.

Does the agent watch live traffic?

No. Savepoint Arc is not a live monitor and does not block traffic. It reads files you export from your IDS, edits rule or configuration files that you approve, and runs your own commands through Shell. Watching the network stays the job of your sensors.

What is network intrusion detection?

It is the monitoring of traffic on a network segment, so alerts describe connections and packets rather than one computer. It sees no traffic. It reads only the alert export from your sensor and helps you decide which alerts were expected.

How does a host based intrusion detection system work?

A host-based IDS watches one machine: its logs, files and processes. Alerts come from changes there. Export them to a file on your PC, and the agent can filter them with grep, read the context with read_file and draft a summary for you to review.

What are intrusion detection and prevention systems?

They are IDS and IPS tools working as a pair. Roles differ. The IDS part raises alerts about suspicious behaviour, and the IPS part can also block it. Savepoint Arc does neither of those jobs, but it can edit the text files around them, such as rules and notes, once you approve each diff.

What is a network based intrusion detection system?

It is an IDS that looks at traffic on a network segment, as opposed to one that looks at a single machine. Many setups use both. The agent works from the alert exports of either kind and never touches the live traffic.

Get started

Try it on one alert export

Download the setup program and open a folder with one alert export. Start with Plan only. Read the plan, switch to Ask first, and check each of the 3 or 4 proposed edits as a diff before you accept it.

Download Savepoint Arc for Windows ↗

More about Savepoint Arc

Every page of this website is available in 20 languages.

What the agent does with rule files ↗ Connecting a cloud, local or built-in model ↗ Step-by-step alert triage guide ↗ Download and checksum steps ↗ All questions and answers ↗ Privacy notice ↗