diff --git a/README.md b/README.md
index 5eb4ce1..9a802c8 100644
--- a/README.md
+++ b/README.md
@@ -4,11 +4,111 @@ Tool to make meta rules from much simpler config files
 
 ## Synopsis
 
-Work-in-progress
+This tool will take simple input like:
 
-Tool currently passes all tests and generates valid config files. Code needs to
-be cleaned up, the installation script needs to be finished and documentation
-improved.
+```
+word
+another 2
+final group LOCAL
+```
+
+and generate complex meta rules to match the first keyword per line. The
+optional number represents a standalone score for just that word while all
+subsequent words represent a keyword group.
+
+If no score is included, the word will only be used for a counter for all words
+in the relevant groups. If no groups are listed, 'LOCAL' is implied. If one or
+more groups is listed, LOCAL must be listed explicitly, otherwise that word
+will not be used in the counter for the other words in that file.
+
+The relatively simple output above will produce the following output:
+
+```
+body    __MC_TEST_ANOTHER_BODY /\banother\b/
+header  __MC_TEST_ANOTHER_SUBJ Subject =~ /\banother\b/
+meta    __MC_TEST_ANOTHER ( __MC_TEST_ANOTHER_BODY || __MC_TEST_ANOTHER_SUBJ )
+
+body    __MC_TEST_WORD_BODY /\bword\b/
+header  __MC_TEST_WORD_SUBJ Subject =~ /\bword\b/
+meta    __MC_TEST_WORD ( __MC_TEST_WORD_BODY || __MC_TEST_WORD_SUBJ )
+
+body    __MC_TEST_FINAL_BODY /\bfinal\b/
+header  __MC_TEST_FINAL_SUBJ Subject =~ /\bfinal\b/
+meta    __MC_TEST_FINAL ( __MC_TEST_FINAL_BODY || __MC_TEST_FINAL_SUBJ )
+
+# LOCAL
+meta    MC_TEST_1 ( __MC_TEST_WORD + __MC_TEST_ANOTHER + __MC_TEST_FINAL ) >= 1
+
+meta    MC_TEST_2 ( __MC_TEST_WORD + __MC_TEST_ANOTHER + __MC_TEST_FINAL ) >= 2
+
+meta    MC_TEST_3 ( __MC_TEST_WORD + __MC_TEST_ANOTHER + __MC_TEST_FINAL ) >= 3
+
+# GROUP
+meta    MC_TEST_GROUP_1 ( __MC_TEST_FINAL ) >= 1
+
+meta    MC_TEST_ANOTHER ( __MC_TEST_ANOTHER )
+
+# SCORES
+describe    MC_TEST_1 Found 1 LOCAL word from MC_TEST
+score       MC_TEST_1 0.01
+
+describe    MC_TEST_2 Found 2 LOCAL words from MC_TEST
+score       MC_TEST_2 0.01
+
+describe    MC_TEST_3 Found 3 LOCAL words from MC_TEST
+score       MC_TEST_3 0.01
+
+describe    MC_TEST_GROUP_1 Found 1 GROUP word from MC_TEST
+score       MC_TEST_GROUP_1 0.01
+
+score       MC_TEST_ANOTHER 2
+```
+
+The initial 'body' and 'header' rules match the keyword in that respective
+message component, and the first 'meta' rules simply indicate if one or then
+other were hit. There are then a series of rules for each available hit count
+for each group. Then there is a list of rules for the individual scoring rules.
+Then, finally, there are all of the score definitions.
+
+There are options to allow scores and rules to be combined in the same file
+or split into separate files.
+
+By default, a different (set of) file(s) will be generated for each input file,
+but it is also possible to combine them into a single file. That same 'global'
+file may already exist if any keywords are assigned to the special 'GLOBAL'
+group.
+
+For more detailed documentation, see inline POD or use perldoc after
+installation:
+
+```
+perldoc Mail::SpamAssassin::KeywordRuleGenerator
+```
+
+## Dependencies
+
+Mail::SpamAssassin - Required to test output and run build tests
+
+This can be installed via CPAN, but it is also likely included in your
+distribution's repositories:
+
+```
+# Debian/Ubuntu
+apt install spamassassin
+# RHEL/Fedora
+dnf install spamassassin
+```
+
+## Installation
+
+Clone the repository, change directory, generate makefile, then make:
+
+```
+git clone https://git.john.me.tz/jpm/Mail-SpamAssassin-KeywordRuleGenerator.git
+cd KeywordRuleGenerator
+./Makefile.PL
+make install
+```
 
 ## Disclaimer