Answers from the documentation; checking which model is on duty
Hi - I answer from the OpenSmartRoute documentation: routing, the API, plans and quotas, self-hosting. Ask away, or open a support ticket if you need a person.
Grounded in the docs - follow a source before acting on it.
dfir - Skill - OpenSmartRoute
Skillv1.0.0
dfir
Digital forensics and incident response - Windows event log analysis, PCAP forensics, filesystem artifact analysis, AD attack detection, and timeline correlation.
Imported from transilienceai/communitytools (skills/dfir/SKILL.md). Install upstream with npx skills add transilienceai/communitytools --skill dfir. Copyright stays with the author.
DFIR
Investigate security incidents by analyzing event logs, network captures, and filesystem artifacts. Detect and reconstruct AD attack chains.
Techniques
Domain
Key Capabilities
Windows Event Logs
EVTX parsing, Event ID correlation, logon tracking, privilege enumeration
Answer formatting: When forensics questions ask for "the value" of a code variable (e.g., PHP $shell), include language-specific string delimiters and terminators (e.g., 'value'; not just value). Check placeholder hints for format clues.
For malicious Office OOXML, inspect more than VBA streams: attackers may split staged Base64 or script content across drawing/object descriptors, shared strings, named cells, and hidden UserForm control captions/values.
When a VBA byte array starts with an fnstenv/pop decoder stub, convert signed integers to raw bytes and test a Shikata-style rolling XOR decode before treating the shellcode as corrupt.
For legacy Excel BIFF/XLS malware, inspect BOUNDSHEET records for hidden or very hidden worksheets and specifically check for Excel 4.0 macro sheets; changing the hidden-state byte or parsing the sheet directly can expose staged strings and flag fragments that never appear in normal workbook views.
For webshell traffic in PCAPs, recover static keys from the uploaded server-side code first, then decrypt operator tasking before chasing later payloads; if a dropped XOR key file is referenced by a shellcode stage, verify where the encoded region actually starts instead of XORing the whole blob from offset zero.
PowerShell stager pattern (in-place reverse + base64 + IEX): when a stage-1 PS script does [array]::Reverse($charArr) followed by FromBase64String("$charArr"), the -join line is often a red herring — string interpolation of a char array uses $OFS=' ' and FromBase64String tolerates whitespace. Reverse the original base64 string (not the joined-with-spaces version) and decode to get stage 2.
Multi-fragment flag exfil: forensic challenges may split a flag across (a) a hardcoded $partN in the malware that is defined but never referenced (often base64'd), and (b) a field of the captured C2 POST body. Decrypt the body with the static AES key from the leaked stage-2 source (PowerShell Encrypt-String puts IV‖ciphertext then base64-wraps); inspect every JSON field for further base64.
All timestamps in UTC — convert from local time zones in pcap/logs. AM/PM trap: 12:XX AM = 00:XX (midnight), 12:XX PM = 12:XX (noon). 12 AM ≠ 01:00.
Parse EVTX with python-evtx (XML namespace: http://schemas.microsoft.com/win/2004/08/events/event)
Use tshark for pcap (not scapy for large files) — filter with -Y display filters
Decompress Win10 prefetch (MAM\x04 header) with dissect.util.compression.lzxpress_huffman
For AES-encrypted ZIPs (compression method 99), use 7z not unzip
Installed into a catalogue, chosen by a router
Install dfir and it becomes one more candidate the router can pick - when it fits.
A listing is a routing target with a manifest: what it does, which domains it covers, what it costs and who publishes it. Once installed it sits beside your own models and tools, is scored like any other candidate for each request, and shows up in the trace when it wins. Ratings come from workspaces that installed it, one per account.
When a provider is down or over quota the request moves on; the trace says it did.
Use it
Copy one of these into your project. Installing also returns the manifest and these snippets.
# after Install: the listing is in your workspace's routing pool - a plan picks it for its slotcurl-s-X POST https://api.opensmartroute.ai/api/v1/route -H 'Authorization: Bearer $OSR_API_KEY' -H 'Content-Type: application/json' -d '{"text": "...", "plan": true}'
Manifest
An Open Capability Manifest: the router reads it to know what this does, what it costs and when to pick it.
transilienceai-communitytools-dfir.ocm.jsonjson
{
"ocm": "1",
"id": "transilienceai-communitytools-dfir",
"kind": "skill",
"name": "dfir",
"description": "Digital forensics and incident response - Windows event log analysis, PCAP forensics, filesystem artifact analysis, AD attack detection, and timeline correlation. Use when investigating security incidents, analyzing Sherlocks, or performing threat hunting on provided evidence files.",
"publisher": "transilienceai",
"version": "1.0.0",
"capabilities": {
"domains": [
"observability",
"coding"
],
"tags": [
"skill-md",
"skills-sh"
],
"languages": [
"en"
]
},
"quality_prior": 0.6,
"examples": [
"Digital forensics and incident response - Windows event log analysis, PCAP forensics, filesystem artifact analysis, AD attack detection, and timeline correlation. Use when investigating security incidents, analyzing Sherlocks, or performing threat hunting on provided evidence files."
],
"primary": false,
"metadata": {
"source": {
"provider": "skills.sh",
"repository": "https://github.com/transilienceai/communitytools",
"path": "skills/dfir/SKILL.md",
"ref": "HEAD",
"url": "https://github.com/transilienceai/communitytools/blob/HEAD/skills/dfir/SKILL.md",
"key": "transilienceai/communitytools/skills/dfir/SKILL.md"
}
},
"instructions": "# DFIR\n\nInvestigate security incidents by analyzing event logs, network captures, and filesystem artifacts. Detect and reconstruct AD attack chains.\n\n## Techniques\n\n| Domain | Key Capabilities |\n|--------|-----------------|\n| **Windows Event Logs** | EVTX parsing, Event ID correlation, logon tracking, privilege enumeration |\n| **Network Forensics** | PCAP analysis, NTLM extraction, LLMNR/NBT-NS poisoning detection, relay identification |\n| **Filesystem Forensics** | MFT parsing, Prefetch analysis, VSS artifact recovery, Linux persistence, timeline reconstruction |\n| **AD Attack Detection** | Kerberoasting, AS-REP roasting, NTDS dump, NTLM relay, credential theft |\n| **Memory Forensics** | Volatility3 analysis: process trees, file extraction, SID resolution, command lines |\n| **Hash Analysis** | NTLMv2 hash construction from pcap, offline cracking validation |\n\n## Workflow\n\n1. **Inventory evidence** — List all artifacts (EVTX, pcap, MFT, prefetch, registry)\n2. **Parse structured data** — EVTX with `python-evtx`, pcap with `tshark`, MFT with `analyzeMFT`\n3. **Identify attack indicators** — Key Event IDs, suspicious traffic patterns, anomalous files\n4. **Correlate across sources** — Match timestamps, IPs, LogonIDs, and process IDs across artifacts\n5. **Reconstruct timeline** — Build chronological attack chain with UTC timestamps\n6. **Answer investigative questions** — Map findings to specific incident response queries\n\n## Tools\n\n```bash\npip install python-evtx windowsprefetch analyzeMFT\nbrew install wireshark p7zip hashcat\n```\n\n| Tool | Purpose |\n|------|---------|\n| `python-evtx` | Parse Windows .evtx files |\n| `tshark` | CLI pcap analysis (NTLM, LLMNR, SMB filters) |\n| `analyzeMFT` | Parse NTFS Master File Table |\n| `windowsprefetch` | Parse Windows prefetch files (Windows host only) |\n| `hashcat` | Hash cracking (NTLMv2 mode 5600, Kerberos mode 13100/18200) |\n| `volatility3` | Memory dump analysis (pstree, filescan, dumpfiles, getsid, cmdline) |\n| `7z` | Extract AES-encrypted evidence ZIPs |\n\n## Quick Reference: Key Event IDs\n\n| Event ID | Log | Indicates |\n|----------|-----|-----------|\n| 4624 | Security | Successful logon (check Type + IP mismatch) |\n| 4768 | Security | TGT request (PreAuthType=0 → AS-REP roast) |\n| 4769 | Security | TGS request (EncType=0x17 → Kerberoast) |\n| 4799 | Security | Group membership enumerated (VSS/ntdsutil) |\n| 5140 | Security | Network share accessed |\n| 7036 | System | Service state change (VSS start → NTDS dump) |\n| 325/326/327 | Application | ESENT database create/detach/close |\n| 330 | Application | ESENT database file info |\n| 3006/3008 | DNS Client Events | DNS query sent/response received (malicious domain lookups) |\n| 106/200 | Task Scheduler | Scheduled task created/executed (persistence via schtasks) |\n\n## Reference\n\n- [windows-event-analysis.md](reference/windows-event-analysis.md) — EVTX parsing patterns and AD attack detection\n- [network-forensics.md](reference/network-forensics.md) — PCAP analysis for NTLM, LLMNR, relay detection\n- [filesystem-forensics.md](reference/filesystem-forensics.md) — MFT, Prefetch, VSS artifact analysis\n\n## Critical Rules\n\n- **Answer formatting**: When forensics questions ask for \"the value\" of a code variable (e.g., PHP `$shell`), include language-specific string delimiters and terminators (e.g., `'value';` not just `value`). Check placeholder hints for format clues.\n- For malicious Office OOXML, inspect more than VBA streams: attackers may split staged Base64 or script content across drawing/object descriptors, shared strings, named cells, and hidden UserForm control captions/values.\n- When a VBA byte array starts with an `fnstenv`/`pop` decoder stub, convert signed integers to raw bytes and test a Shikata-style rolling XOR decode before treating the shellcode as corrupt.\n- For legacy Excel BIFF/XLS malware, inspect `BOUNDSHEET` records for `hidden` or `very hidden` worksheets and specifically check for Excel 4.0 macro sheets; changing the hidden-state byte or parsing the sheet directly can expose staged strings and flag fragments that never appear in normal workbook views.\n- For webshell traffic in PCAPs, recover static keys from the uploaded server-side code first, then decrypt operator tasking before chasing later payloads; if a dropped XOR key file is referenced by a shellcode stage, verify where the encoded region actually starts instead of XORing the whole blob from offset zero.\n- **PowerShell stager pattern** (in-place reverse + base64 + IEX): when a stage-1 PS script does `[array]::Reverse($charArr)` followed by `FromBase64String(\"$charArr\")`, the `-join` line is often a red herring — string interpolation of a char array uses `$OFS=' '` and `FromBase64String` tolerates whitespace. Reverse the original base64 string (not the joined-with-spaces version) and decode to get stage 2.\n- **Multi-fragment flag exfil**: forensic challenges may split a flag across (a) a hardcoded `$partN` in the malware that is *defined but never referenced* (often base64'd), and (b) a field of the captured C2 POST body. Decrypt the body with the static AES key from the leaked stage-2 source (PowerShell `Encrypt-String` puts `IV‖ciphertext` then base64-wraps); inspect every JSON field for further base64.\n- All timestamps in **UTC** — convert from local time zones in pcap/logs. **AM/PM trap**: 12:XX AM = 00:XX (midnight), 12:XX PM = 12:XX (noon). 12 AM ≠ 01:00.\n- Parse EVTX with `python-evtx` (XML namespace: `http://schemas.microsoft.com/win/2004/08/events/event`)\n- Use `tshark` for pcap (not scapy for large files) — filter with `-Y` display filters\n- Decompress Win10 prefetch (MAM\\x04 header) with `dissect.util.compression.lzxpress_huffman`\n- For AES-encrypted ZIPs (compression method 99), use `7z` not `unzip`",
"cost": {
"context_tokens": 1443
}
}
Fetch it by URL: GET /api/v1/registry/transilienceai-communitytools-dfir/manifest?version=1.0.0
Reviews
Star ratings from people who tried it. One review per account; edit yours any time.
No reviews yet. Install it, try it, and be the first to rate it.