stegg-cli

# ST3GG CLI Subprocess steganography CLI. Output stays out of LLM context — call via `python3 stegg_cli.py <command>`. All output is JSON. **This is the primary interface for steganography operations.** Prefer this over the MCP server for routine use (zero context cost). Fall back to MCP tools when you need results inline. ## Commands ``` encode -i IMAGE [-t TEXT|-f FILE] [-o OUT] [--channels C] [--bits N] [--password P] decode -i IMAGE [-o OUT] [--no-auto] [--channels C] [--bits N] [--password P] analyze IMAGE [--full] detect IMAGE capacity IMAGE [--channels C] [--bits N] inject-chunk -i IMAGE -o OUT --text TEXT [--type tEXt] [--keyword Comment] read-chunks IMAGE inject-exif -i IMAGE -o OUT [--comment C] [--author A] [--custom-fields JSON] inject-name [--template T] [--channels C] [--count N] templates analysis-tool IMAGE ACTION list-tools crypto-status ``` ## Quick Reference ### Encode + Verify + Decode ```bash # Hide text python3 stegg_cli.py encode -i carrier.png -t "secret message" -o stegged.png # Always verify roundtrip after encoding python3 stegg_cli.py decode -i stegged.png # Hide file with encryption python3 stegg_cli.py encode -i carrier.png -f payload.bin -o stegged.png --password s3cret python3 stegg_cli.py decode -i stegged.png --password s3cret # Decode with manual config (non-interleaved strategies) python3 stegg_cli.py decode -i stegged.png --no-auto --strategy sequential ``` ### Analyze + Detect ```bash # Quick header check python3 stegg_cli.py detect suspect.png # Chi-square anomaly analysis python3 stegg_cli.py analyze suspect.png # Full 264-function sweep (PNG only) python3 stegg_cli.py analyze suspect.png --full # Specific analysis function python3 stegg_cli.py analysis-tool suspect.png rs_analysis python3 stegg_cli.py analysis-tool suspect.png sample_pairs_analysis ``` ### Metadata Injection ```bash # PNG chunk python3 stegg_cli.py inject-chunk -i image.png -o out.png --text "hidden metadata" # Private chunk type python3 stegg_cli.py inject-chunk -i image.png -o out.png --type stEg --text "private" # EXIF fields python3 stegg_cli.py inject-exif -i image.png -o out.png --comment "payload" --author "red-team" # Read chunks python3 stegg_cli.py read-chunks image.png ``` ### AI Red Team ```bash # Generate injection filenames python3 stegg_cli.py inject-name --template claude_decoder --channels R --count 3 # List jailbreak templates python3 stegg_cli.py templates ``` ## Key Constraints - **Use `interleaved` strategy** (default) — only strategy with auto-detect on decode - `spread` and `randomized` have upstream decode bugs - `sequential` works but needs `--no-auto` on decode - All output is JSON; errors: `{"error": "..."}` with exit code 1 - **Always check for `error` key** in JSON output before proceeding in multi-step operations - **Always verify encode with decode** before distributing stegged images ## Installation ```bash pip install stegg # CLI is available as stegg-cli, or: python3 /path/to/st3gg/stegg_cli.py <command> ```