From image-annotation
Privacy redaction — pixelate (default), blur, or solid-fill rectangular regions of an image to hide sensitive content (faces, names, tokens, IPs, PII). Originals are never modified; output goes to <input_dir>/redacted/<stem>_redacted<ext>. EXIF is stripped on output.
npx claudepluginhub danielrosehill/claude-code-plugins --plugin image-annotationThis skill uses the workspace's default tool permissions.
Hide sensitive regions of an image. Use this rather than `annotate` when the goal is privacy — defaults are tuned to make the redaction non-recoverable.
Guides Next.js Cache Components and Partial Prerendering (PPR): 'use cache' directives, cacheLife(), cacheTag(), revalidateTag() for caching, invalidation, static/dynamic optimization. Auto-activates on cacheComponents: true.
Guides building MCP servers enabling LLMs to interact with external services via tools. Covers best practices, TypeScript/Node (MCP SDK), Python (FastMCP).
Share bugs, ideas, or general feedback.
Hide sensitive regions of an image. Use this rather than annotate when the goal is privacy — defaults are tuned to make the redaction non-recoverable.
This skill never overwrites the original. Output goes to <input_dir>/redacted/<stem>_redacted<ext> by default, or to --output PATH if supplied. There is no in-place mode for redaction — keep the source intact (or delete it yourself once you've verified the redacted copy).
bash "$CLAUDE_PLUGIN_ROOT/scripts/redact.sh" \
--input "$INPUT" \
[--output "$OUTPUT"] \
[--method pixelate|blur|solid] \
--region X,Y,W,H [--region X,Y,W,H ...]
Gaussian blur on text is reversible with deconvolution; coarse pixelation isn't. The script picks a block size of min(W,H)/8 (floor 16) so even a small region collapses to ≤8 cells per side.
The script appends -strip to the ImageMagick pipeline so GPS, original timestamp, and device metadata don't survive into the redacted output.