cognium-ai 2.5.7 → 2.5.9

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (2) hide show
  1. package/README.md +16 -12
  2. package/package.json +2 -2
package/README.md CHANGED
@@ -12,7 +12,7 @@ npm install -g cognium-ai
12
12
 
13
13
  ```bash
14
14
  cognium-ai scan <path> # Scan for security vulnerabilities (LLM-enhanced)
15
- cognium-ai trust <path> # Trust score across 30+ passes (supply chain, security, AI safety, compliance)
15
+ cognium-ai trust <path> # Trust score across 36 passes (supply chain, security, AI safety, compliance)
16
16
  cognium-ai quality <path> # Quality score (complexity, tests, docs, maintainability, performance)
17
17
  cognium-ai spec-diff <path> # Spec-gap analysis (code vs Specifica spec alignment)
18
18
  cognium-ai generate-spec <path> # Generate Specifica spec from code
@@ -34,15 +34,18 @@ are hidden from the headline `--help` until they soak.
34
34
  cognium-ai scan src/ # LLM-enhanced scan (default)
35
35
  cognium-ai scan src/ --no-llm # Static-only (no LLM)
36
36
  cognium-ai scan src/ -f json -o results.json # JSON output to file
37
+ cognium-ai scan src/ -f sarif -o results.sarif # SARIF 2.1.0 (GitHub code-scanning)
37
38
  cognium-ai scan src/ --severity high # High+ severity only
38
39
  cognium-ai scan src/ --exclude-tests # Skip test files
39
40
  cognium-ai scan src/ --threads 20 # Custom parallelism
40
41
  cognium-ai scan src/ -x '**/vendor/**' # Exclude paths
42
+ cognium-ai scan src/ --llm-timeout 180 # Raise per-call LLM timeout (slow local models)
41
43
  cognium-ai scan src/ --exit-code # Exit 1 on findings (CI)
42
44
  ```
43
45
 
44
- For SARIF output, use `cognium-ai trust` (`-f sarif -o trust.sarif`),
45
- which produces SARIF 2.1.0 against all trust passes.
46
+ `cognium-ai trust src/ -f sarif -o trust.sarif` produces SARIF 2.1.0
47
+ across all 36 trust passes (richer than `scan -f sarif`, which is
48
+ scoped to OWASP Top 10 findings only).
46
49
 
47
50
  ## LLM Configuration
48
51
 
@@ -65,9 +68,9 @@ export LLM_ENRICHMENT_MODEL=cognium/gpt-oss-120b
65
68
  |------|-------------|---------|
66
69
  | `--llm-base-url <url>` | LLM API base URL (OpenAI-compatible) | `http://localhost:4000/v1` |
67
70
  | `--llm-api-key <key>` | LLM API key | `LLM_API_KEY` env var |
68
- | `--llm-model <model>` | LLM model name | `cognium/gpt-oss-120b` |
71
+ | `--llm-model <model>` | LLM model name (universal — applies to all phases) | `cognium/gpt-oss-120b` |
72
+ | `--llm-timeout <seconds>` | Per-call LLM timeout. Raise for slow local models (e.g. `--llm-timeout 180`); lower for fail-fast iteration. | `60` |
69
73
  | `--no-llm` | Disable LLM, use static analysis only | LLM enabled by default |
70
- | `--no-llm-discovery` | Disable LLM discovery mode | discovery enabled by default |
71
74
 
72
75
  ### Provider Examples
73
76
 
@@ -141,15 +144,16 @@ jobs:
141
144
 
142
145
  | Benchmark | Score |
143
146
  |-----------|-------|
144
- | OWASP Benchmark (Java, 1415 tests) | +100% |
145
- | Juliet Test Suite (156 tests) | +100% |
147
+ | OWASP Benchmark (Java) | 100% (1415/1415) |
148
+ | Juliet Test Suite (14 CWEs) | 100% (243/243) |
146
149
  | SecuriBench Micro | 97.7% TPR, 6.7% FPR |
147
- | CWE-Bench-Java (120 CVEs) | 42.5% static, 81.7% +LLM Discovery |
148
- | NodeJS Synthetic (25 tests) | 100% TPR |
149
- | CWE-Bench-Rust (30 tests) | 77.8% TPR, 0% FPR |
150
- | Bash Synthetic (31 tests) | 68.2% TPR, 0% FPR |
150
+ | CWE-Bench-Java (120 CVEs) | 50.8% static (61/120), **86.7% +LLM Discovery** (104/120, Claude Opus) |
151
+ | OWASP NodeGoat / Juice Shop / DVJA | 100% |
152
+ | NodeJS Synthetic (25 tests) | 92.9% Score (96.2% TPR, 11.1% FPR) |
153
+ | CWE-Bench-Rust (30 tests) | 94.4% TPR, 0% FPR |
154
+ | Bash Synthetic (31 tests) | 100% TPR, 0% FPR |
151
155
 
152
- CWE-Bench-Java reference: CodeQL 22.5%, IRIS+GPT-4 45.8%.
156
+ CWE-Bench-Java reference: **CodeQL 22.5%**, IRIS+GPT-4 45.8% — cognium-ai with Claude Opus is **2.85× CodeQL**.
153
157
 
154
158
  ## License
155
159
 
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "cognium-ai",
3
- "version": "2.5.7",
3
+ "version": "2.5.9",
4
4
  "description": "AI-powered static analysis CLI with LLM-enhanced vulnerability detection",
5
5
  "main": "dist/index.js",
6
6
  "types": "dist/index.d.ts",
@@ -42,7 +42,7 @@
42
42
  ],
43
43
  "dependencies": {
44
44
  "circle-ir": "^3.19.5",
45
- "circle-ir-ai": "^2.5.6",
45
+ "circle-ir-ai": "^2.5.7",
46
46
  "commander": "^14.0.3",
47
47
  "minimatch": "^10.2.5"
48
48
  },