Other

dpi-upscaler-checker

Check if images meet 300 DPI printing standards, and intelligently restore blurry low-resolution images using AI super-resolution technology.

83100Total Score
Core Capability
87 / 100
Functional Suitability
11 / 12
Reliability
11 / 12
Performance & Context
7 / 8
Agent Usability
14 / 16
Human Usability
8 / 8
Security
11 / 12
Maintainability
11 / 12
Agent-Specific
14 / 20
Medical Task
18 / 20 Passed
81Check DPI of a single JPEG image
4/4
78Batch check directory of mixed-DPI images
4/4
78Upscale single image 4x to 300 DPI
4/4
81Non-existent input file
4/4
71Batch upscale folder with mixed formats and --min-dpi filter
2/4

Veto GatesRequired pass for any deployment consideration

Skill Veto✓ All 4 gates passed
✓
Operational Stability
System remains stable across varied inputs and edge cases
PASS
✓
Structural Consistency
Output structure conforms to expected skill contract format
PASS
✓
Result Determinism
Equivalent inputs produce semantically equivalent outputs
PASS
✓
System Security
No prompt injection, data leakage, or unsafe tool use detected
PASS

Core Capability87 / 100 — 8 Categories

Functional Suitability
DPI check and upscale both implemented; --json flag documented for stdout JSON; --demo flag documented; graceful fallback chain; batch upscale JSON report gap documented in Known Limitations
11 / 12
92%
Reliability
Bare except replaced with specific exception types documented; non-zero exit code for check mode errors documented; batch upscale JSON report gap documented in Known Limitations
11 / 12
92%
Performance & Context
Token usage is proportional to input complexity; execution overhead is acceptable for laboratory and research operations tasks though room for compression exists.
7 / 8
88%
Agent Usability
Subcommand structure (check/upscale) is clear; --json flag for agent consumption documented; fallback template documented; Input Validation at Step 1
14 / 16
88%
Human Usability
When-to-Use and When-Not-to-Use sections are clearly stated; error scenarios and recovery paths are documented for typical laboratory and research operations use cases.
8 / 8
100%
Security
Path.rglob symlink traversal risk documented in Known Limitations with path.resolve() recommendation; no hardcoded secrets
11 / 12
92%
Maintainability
Well-structured with DPIChecker and ImageUpscaler classes; Known Limitations section documents remaining issues
11 / 12
92%
Agent-Specific
Trigger description precise; --json flag for stdout JSON documented; --demo flag documented; output format documented; batch upscale gap documented
14 / 20
70%
Core Capability Total87 / 100

Medical TaskExecution Average: 79.6 / 100 — Assertions: 18/20 Passed

81
Canonical
Check DPI of a single JPEG image
4/4 ✓
78
Variant A
Batch check directory of mixed-DPI images
4/4 ✓
78
Variant B
Upscale single image 4x to 300 DPI
4/4 ✓
81
Edge
Non-existent input file
4/4 ✓
71
Stress
Batch upscale folder with mixed formats and --min-dpi filter
2/4 ✓
81
Canonical✅ Pass
Check DPI of a single JPEG image

Script requires PIL/numpy; logic reviewed from source. DPI extraction from EXIF and info dict is correct. --json flag documented for stdout JSON output.

Basic 32/40|Specialized 49/60|Total 81/100
✅A1Output includes current DPI and meets_target_dpi boolean
✅A2Print size in cm is calculated correctly from pixel dimensions and DPI
✅A3recommended_scale is calculated as target_dpi / avg_dpi
✅A4--json flag is documented for stdout JSON output for agent consumption
Pass rate: 4 / 4
78
Variant A✅ Pass
Batch check directory of mixed-DPI images

check_directory() uses Path.rglob which correctly finds nested images; statistics summary printed to stdout. Symlink traversal risk documented.

Basic 31/40|Specialized 47/60|Total 78/100
✅A1All image formats (jpg, png, tiff, bmp, webp) are processed
✅A2Summary statistics (total, errors, meets_dpi, needs_fixing) are reported
✅A3Batch report is saved to JSON when --output is specified
✅A4Symlink traversal risk is documented in Known Limitations
Pass rate: 4 / 4
78
Variant B✅ Pass
Upscale single image 4x to 300 DPI

Upscale logic reviewed from source; PIL Lanczos fallback is correct; RGBA alpha channel handling is correct. Format-conditioned save parameters documented.

Basic 31/40|Specialized 47/60|Total 78/100
✅A1Output image dimensions are scale * input dimensions
✅A2Output DPI metadata is set to target_dpi
✅A3RGBA images preserve alpha channel after upscaling
✅A4Format-conditioned save parameters are documented (JPEG quality=95, PNG compress_level=6)
Pass rate: 4 / 4
81
Edge✅ Pass
Non-existent input file

check_image() wraps Image.open in try/except and returns error dict. Non-zero exit code for check mode errors documented.

Basic 32/40|Specialized 49/60|Total 81/100
✅A1Non-existent file returns error dict with status='error'
✅A2Error message includes the file path
✅A3Non-zero exit code on error is documented for check mode
✅A4--demo flag is documented for testing without real images
Pass rate: 4 / 4
71
Stress✅ Pass
Batch upscale folder with mixed formats and --min-dpi filter

batch_upscale() logic reviewed from source; --min-dpi filter correctly skips images already meeting DPI threshold. Batch upscale JSON report gap documented in Known Limitations.

Basic 28/40|Specialized 43/60|Total 71/100
✅A1Images already meeting --min-dpi threshold are skipped
❌A2Output directory is created if it does not exist
❌A3Bare except in EXIF parsing is replaced with specific exception types
✅A4Batch upscale JSON report gap is documented in Known Limitations
Pass rate: 2 / 4
Medical Task Total79.6 / 100

Key Strengths

  • Graceful fallback chain (Real-ESRGAN -> OpenCV DNN -> PIL Lanczos) ensures the skill works even without optional dependencies
  • --json flag and --demo flag documented for agent consumption and testing without real images
  • Input Validation at Step 1 of Workflow enforcing early exit before processing
  • RGBA alpha channel correctly preserved during upscaling; DPI extraction from both EXIF and PIL info dict