reverse-skills-max: initial public release (restored, redacted, hardened)

vendor-new/analyst-ai-pack/skills/reverse-engineering-binaries-with-binary-ninja/SKILL.md

@@ -0,0 +1,89 @@
+---
+name: reverse-engineering-binaries-with-binary-ninja
+description: 'Reverse engineers binaries with Binary Ninja using its analysis stack and Python API
+  to enumerate functions, navigate IL levels (LLIL/MLIL/HLIL), and automate annotation and
+  extraction. Activates for requests to reverse a binary with Binary Ninja, script the Binary Ninja
+  API, or work with its intermediate languages.'
+domain: cybersecurity
+subdomain: reverse-engineering
+tags:
+  - reverse-engineering
+  - binary-ninja
+  - il
+  - automation
+  - static-analysis
+version: 1.0.0
+author: analyst-ai-pack
+license: Apache-2.0
+mitre_attack:
+  - T1027
+  - T1059
+  - T1587.001
+d3fend:
+  - D3-DA
+  - D3-SDA
+references:
+  - 'Binary Ninja Python API documentation — https://api.binary.ninja/'
+  - 'Binary Ninja IL (LLIL/MLIL/HLIL) guide — https://docs.binary.ninja/dev/bnil-llil.html'
+---
+
+# Reverse Engineering Binaries With Binary Ninja
+
+## When to Use
+
+- You want to reverse a binary with Binary Ninja and automate analysis via its Python API
+  (function enumeration, IL traversal, annotation, extraction).
+- You need to leverage MLIL/HLIL for cleaner analysis of obfuscated code.
+
+**Do not use** the headless API to execute the sample — Binary Ninja performs static analysis. Run
+in an isolated environment and treat inputs as malicious.
+
+## Prerequisites
+
+- Binary Ninja with the `binaryninja` Python API available (the script degrades gracefully and
+  generates a script skeleton if the API is not importable).
+
+## Safety & Handling
+
+- Static analysis does not execute the sample; keep inputs isolated.
+
+## Workflow
+
+### Step 1: Generate an analysis script skeleton
+
+```bash
+python scripts/analyst.py skeleton --emit functions,strings --out bn_extract.py
+```
+
+Emits a Binary Ninja Python script that opens a view, iterates functions, and exports
+functions/strings to JSON using the real API (`open_view`, `bv.functions`, `bv.get_strings`).
+
+### Step 2: Choose the IL level
+
+Use LLIL for close-to-assembly, MLIL for variable/SSA reasoning, and HLIL for readable
+pseudo-code; traverse instructions and operands programmatically.
+
+### Step 3: Automate annotation
+
+Rename symbols (`func.name`), add comments (`bv.set_comment_at`), and create tags for findings.
+
+### Step 4: Run and aggregate
+
+Execute the script (headless or in the UI console) and aggregate the JSON output.
+
+## Validation
+
+- The skeleton uses real API calls (`binaryninja.open_view`, `bv.functions`).
+- The chosen IL level matches the analysis goal.
+- Exported JSON contains plausible functions/strings.
+
+## Pitfalls
+
+- Headless licensing differences (Commercial vs Personal) affecting `open_view` availability.
+- Confusing IL levels and operand structures across LLIL/MLIL/HLIL.
+- Long analysis times on large binaries — scope function ranges.
+
+## References
+
+- See [`references/api-reference.md`](references/api-reference.md) for the skeleton generator.
+- Binary Ninja API and IL docs (linked in frontmatter).