tersify 0.5.0

Universal LLM context compressor — pipe anything, get token-optimized output
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87

name: 'tersify'
description: 'Compress code and context before sending to LLMs — saves 30–54% of tokens.'
author: 'rustkit-ai'

branding:
  icon: 'zap'
  color: 'orange'

inputs:
  path:
    description: 'File or directory to compress (default: current directory)'
    required: false
    default: '.'
  ast:
    description: 'Extract function signatures only — stub all bodies with { /* ... */ }'
    required: false
    default: 'false'
  strip-docs:
    description: 'Also remove doc comments (///, //!, /** */, Python docstrings)'
    required: false
    default: 'false'
  output-file:
    description: 'Write compressed output to this file path instead of the step output'
    required: false
    default: ''
  verbose:
    description: 'Print token savings summary'
    required: false
    default: 'true'

outputs:
  compressed:
    description: 'The compressed content (only set when output-file is not specified)'
    value: ${{ steps.compress.outputs.compressed }}
  tokens-before:
    description: 'Token count before compression'
    value: ${{ steps.compress.outputs.tokens_before }}
  tokens-after:
    description: 'Token count after compression'
    value: ${{ steps.compress.outputs.tokens_after }}
  savings-pct:
    description: 'Percentage of tokens saved (e.g. "35%")'
    value: ${{ steps.compress.outputs.savings_pct }}

runs:
  using: 'composite'
  steps:
    - name: Install tersify
      shell: bash
      run: |
        if ! command -v tersify &>/dev/null; then
          curl -fsSL https://raw.githubusercontent.com/rustkit-ai/tersify/main/install.sh | bash
          echo "$HOME/.local/bin" >> "$GITHUB_PATH"
        fi

    - name: Compress
      id: compress
      shell: bash
      run: |
        FLAGS=""
        if [ "${{ inputs.ast }}" = "true" ];        then FLAGS="$FLAGS --ast"; fi
        if [ "${{ inputs.strip-docs }}" = "true" ]; then FLAGS="$FLAGS --strip-docs"; fi
        if [ "${{ inputs.verbose }}" = "true" ];    then FLAGS="$FLAGS --verbose"; fi

        # Run tersify; capture stdout (compressed) and stderr (savings summary)
        COMPRESSED=$(tersify $FLAGS "${{ inputs.path }}" 2>/tmp/_tersify_verbose)
        SUMMARY=$(cat /tmp/_tersify_verbose || true)

        # Parse token counts from summary line: "[tersify] 5439 → 3559 tokens  (35% saved)"
        TOKENS_BEFORE=$(echo "$SUMMARY" | grep -oP '(?<=\[tersify\] )\d+' || echo "0")
        TOKENS_AFTER=$(echo  "$SUMMARY" | grep -oP '(?<=→ )\d+(?= tokens)' || echo "0")
        SAVINGS_PCT=$(echo   "$SUMMARY" | grep -oP '\d+(?=% saved)'        || echo "0")

        if [ -n "${{ inputs.output-file }}" ]; then
          echo "$COMPRESSED" > "${{ inputs.output-file }}"
          echo "compressed=" >> "$GITHUB_OUTPUT"
        else
          {
            echo "compressed<<TERSIFY_EOF"
            echo "$COMPRESSED"
            echo "TERSIFY_EOF"
          } >> "$GITHUB_OUTPUT"
        fi

        echo "tokens_before=$TOKENS_BEFORE" >> "$GITHUB_OUTPUT"
        echo "tokens_after=$TOKENS_AFTER"   >> "$GITHUB_OUTPUT"
        echo "savings_pct=${SAVINGS_PCT}%"  >> "$GITHUB_OUTPUT"