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
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140

# MCP Server

tersify ships a built-in [MCP](https://modelcontextprotocol.io/) server for agent pipelines. It exposes three tools over JSON-RPC 2.0 (stdio transport, protocol `2024-11-05`).

---

## Start the server

```bash
tersify mcp
```

The server reads JSON-RPC requests from stdin and writes responses to stdout.

---

## Register with Claude Code

```bash
claude mcp add tersify -- tersify mcp
```

After this, Claude Code can call `compress`, `count_tokens`, and `estimate_cost` as native tools — no more shelling out manually.

---

## Tools

### `compress`

Compress text to reduce LLM token usage.

**Input:**

| Parameter | Type | Required | Description |
|---|---|:---:|---|
| `text` | string | ✓ | The text to compress |
| `content_type` | string | — | Force type: `rust` `python` `js` `ts` `go` `ruby` `java` `c` `json` `logs` `diff` `text` — omit for auto-detection |
| `ast` | boolean | — | Extract signatures only (default: false) |
| `smart` | boolean | — | Near-duplicate deduplication (default: false) |
| `budget` | integer | — | Hard token cap |

**Output:**

```json
{
  "content": [{ "type": "text", "text": "<compressed output>" }],
  "meta": {
    "tokens_before": 384,
    "tokens_after": 228,
    "saved_pct": "41%",
    "content_type": "rust"
  }
}
```

---

### `count_tokens`

Count the approximate number of LLM tokens in a string (~4 chars/token).

**Input:**

| Parameter | Type | Required | Description |
|---|---|:---:|---|
| `text` | string | ✓ | The text to count |

**Output:**

```json
{
  "content": [{ "type": "text", "text": "96 tokens" }],
  "meta": { "count": 96 }
}
```

---

### `estimate_cost`

Show API cost before and after compression across all major LLM providers.

**Input:**

| Parameter | Type | Required | Description |
|---|---|:---:|---|
| `text` | string | ✓ | The text to estimate |
| `content_type` | string | — | Force type (same values as compress) |
| `model` | string | — | Filter models by name (e.g. `"claude"`, `"gpt-4o"`) |

**Output:**

```json
{
  "content": [{ "type": "text", "text": "384 → 228 tokens (41% saved)\n\nModel  ..." }],
  "meta": {
    "tokens_before": 384,
    "tokens_after": 228,
    "saved_pct": "41%",
    "models": [
      {
        "model": "claude-sonnet-4.6",
        "cost_raw_usd": 0.001152,
        "cost_compressed_usd": 0.000684,
        "cost_saved_usd": 0.000468
      }
    ]
  }
}
```

---

## Protocol details

- Transport: **stdio** (JSON-RPC 2.0, one message per line)
- Protocol version: `2024-11-05`
- Supported methods: `initialize`, `tools/list`, `tools/call`
- Notifications (no `id`): silently ignored

---

## Use from another MCP client

Any client that speaks MCP stdio works. Spawn the process and communicate over its stdin/stdout:

```
Process: tersify mcp
  stdin  ← your JSON-RPC requests
  stdout → tersify responses
  stderr → tersify logs (prefix: [tersify mcp])
```

Example initialize handshake:

```json
→ {"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{}}}
← {"jsonrpc":"2.0","id":1,"result":{"protocolVersion":"2024-11-05","capabilities":{"tools":{}},"serverInfo":{"name":"tersify","version":"0.3.0"}}}
```