Skip to content
cccskills
FreePublished on GitHub

writing-yara-rules-from-reversed-code

Turns reverse-engineering findings into durable YARA detections: selecting stable code constructs and constants over volatile strings, extracting opcode/byte patterns with wildcards, and validating rules for low false positives. Activates for requests to write a YARA rule from reversed code, create a detection signature, or convert RE findings into hunting rules.

The English summary is being prepared. Showing the original description.

See how to install

Files included (4)

  • SKILL.md3.3 KB
  • LICENSE340 B
  • references/api-reference.md1.2 KB
  • scripts/analyst.py2.1 KB

SKILL.md (original)

Review the instructions the agent will receive before you install.

Writing YARA Rules from Reversed Code

When to Use

  • You finished reversing a sample/family and want a detection that survives recompilation and cosmetic changes.
  • You need byte-pattern signatures from decryptors, API-hash constants, or unique algorithms rather than fragile strings.
  • You are converting RE notes into hunting/scanning rules for a corpus.

Do not use volatile artifacts (file paths, mutex names that change per build, packer stubs shared across unrelated families) as your primary anchor — they cause drift and false hits.

Prerequisites

  • The yara engine (and ideally yara-python) for testing.
  • A small corpus: target samples (true positives) and clean/unrelated files (false-positive control).

Workflow

Step 1: Choose stable anchors

Prefer, in order: a unique algorithm's opcode sequence (decryptor, hashing loop), embedded magic constants (API hashes, XOR keys, S-box), then distinctive strings only if intrinsic.

Step 2: Extract byte patterns with wildcards

Pull the relevant opcodes and wildcard volatile operands (addresses, immediates) so the rule survives relocation/recompilation:

$dec = { 8A 04 ?? 34 ?? 88 04 ?? 41 3B ?? 7? ?? }   ; xor-decrypt loop, regs/disp wildcarded

Step 3: Assemble the rule

Combine 2–3 independent anchors with a condition requiring enough of them, plus a cheap prefilter (file size, PE magic) to keep scanning fast:

python scripts/analyst.py scaffold --name family_xyz --hash 0xABCDEF12

Step 4: Validate against the corpus

Run the rule across true positives and the clean control set; require all TPs match and zero FPs on the control.

yara -r rules/family_xyz.yar ./corpus

Step 5: Tune and document

Adjust thresholds, add meta (author, date, reference, hash), and record which construct each string anchors so future analysts can maintain it.

Validation

  • Rule matches all intended samples and produces zero hits on the clean control set.
  • Anchors map to intrinsic code/constants, not build-specific noise.
  • meta documents source samples and the reasoning for each pattern.

Pitfalls

  • Wildcarding too little (brittle) or too much (false positives); wildcard only volatile bytes.
  • Anchoring on shared library/packer code present in unrelated binaries.
  • Skipping the clean-corpus test, shipping a noisy rule.

References

Reviews

No reviews yet. Tell others what you thought.

Comments

No login needed. Share how it worked for you, ask a question, or note where it did not run.

No comments yet. Be the first to write one.

Comments are published right away. Do not include personal information such as email addresses or phone numbers. Up to 2 URLs.

0 / 1,000 characters

This form is protected by reCAPTCHA, and the Google Privacy Policy and Terms of Service apply.

More skills from this repository

Reverses Android malware: unpacking APKs, decompiling DEX bytecode to readable Java, auditing the manifest for abused permissions and components, and locating dynamically loaded or native payloads. Activates for requests to analyze an APK, decompile DEX, or investigate a suspicious Android app.

The English summary is being prepared. Showing the original description.

meltedinhex/analyst-ai-pack22Updated Jul 7, 2026

Analyzes API call traces from a sandbox or API monitor (JSON) to group calls by category, reconstruct high-level behaviors (process injection, file drops, network, crypto), and flag suspicious call sequences. Activates for requests to analyze an API trace, interpret sandbox API logs, or identify behavior from Win32/Native API calls.

The English summary is being prepared. Showing the original description.

meltedinhex/analyst-ai-pack22Updated Jul 7, 2026

Analyzes Windows Authenticode signatures on PE files: checking for a signature, reading the signer certificate chain, detecting revoked/expired/stolen certs, and recognizing signature-stripping and catalog-signing abuse. Activates for requests to analyze a code signature, verify Authenticode, or assess signer trust on a PE.

The English summary is being prepared. Showing the original description.

meltedinhex/analyst-ai-pack22Updated Jul 7, 2026

Analyzes banking trojan webinject configurations to extract targeted institutions, injected JavaScript/HTML, and data-theft hooks, mapping the fraud workflow. Activates for requests to analyze banking trojan webinjects, parse a webinject config, or map targeted banks and credential-theft injections.

The English summary is being prepared. Showing the original description.

meltedinhex/analyst-ai-pack22Updated Jul 7, 2026

Analyzes Python-based malware packaged as PyInstaller/py2exe executables by detecting the packer, locating the embedded archive, and identifying bundled .pyc modules for extraction and decompilation. Activates for requests to analyze compiled Python malware, unpack a PyInstaller binary, or extract pyc modules from a frozen Python executable.

The English summary is being prepared. Showing the original description.

meltedinhex/analyst-ai-pack22Updated Jul 7, 2026

Analyzes cryptojacking/coinminer malware by extracting mining pool endpoints (stratum), wallet addresses, algorithm and miner identifiers (XMRig and forks), and persistence or resource-control settings from static strings and config. Activates for requests to analyze a cryptominer, extract mining pool and wallet IOCs, or identify coinminer configuration.

The English summary is being prepared. Showing the original description.

meltedinhex/analyst-ai-pack22Updated Jul 7, 2026

See all skills by meltedinhex

Report a problem with this skill