abmind/openclaw.plugin.json at dev · aksika/abmind · GitHub

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
{
  "id": "abmind",
  "name": "abmind Memory + Context Engine",
  "description": "Long-term memory, context compaction, sleep cycles, recall — SQLite-backed, FTS5 + trigram + embeddings.",
  "kind": "memory",
  "entry": "./dist/src/openclaw-plugin/index.js",
  "configSchema": {
    "type": "object",
    "properties": {
      "abmlVersion": {
        "type": "string",
        "enum": ["plain", "v0", "v1"],
        "default": "plain",
        "description": "Rendering format for memory injection. 'plain' = English prose, 'v0' = legacy ABM-L, 'v1' = slot-based ABM-L."
      },
      "compactionThreshold": {
        "type": "number",
        "minimum": 0,
        "maximum": 1,
        "default": 0.5,
        "description": "Fraction of context window at which compaction fires."
      },
      "sleepEnabled": {
        "type": "boolean",
        "default": true,
        "description": "Enable nightly sleep pipeline (extraction, consolidation, cleanup)."
      },
      "stateDir": {
        "type": "string",
        "description": "Override the abmind data directory. Defaults to ABMIND_HOME env or ~/.abmind."
      },
      "autoRecall": {
        "type": "boolean",
        "default": false,
        "description": "Inject top-relevant memories into context before every agent turn. Opt-in: adds ~10-200ms per turn."
      },
      "autoCapture": {
        "type": "boolean",
        "default": false,
        "description": "Capture user messages for memory extraction after every agent turn. Opt-in: uses existing sleep pipeline for dedupe + classify."
      },
      "autoRecallMaxResults": {
        "type": "integer",
        "minimum": 1,
        "maximum": 10,
        "default": 3,
        "description": "Max memories injected per turn when autoRecall is enabled."
      },
      "autoRecallMinScore": {
        "type": "number",
        "minimum": 0,
        "maximum": 1,
        "default": 0.3,
        "description": "Minimum relevance score for a memory to be injected. Below this, the memory is dropped."
      }
    }
  },
  "uiHints": {
    "sleepEnabled": {
      "label": "Enable sleep pipeline",
      "help": "abmind runs its own sleep/consolidation independently of OpenClaw's dreaming. This toggles whether it runs at all."
    },
    "autoRecall": {
      "label": "Auto-recall memories on every turn",
      "help": "Adds top-relevant memories to context before the agent runs. Costs ~10-50ms (FTS+trigram) to ~200ms (with embeddings)."
    },
    "autoCapture": {
      "label": "Auto-capture user messages",
      "help": "Stores user prompts for later memory extraction by the sleep pipeline. Does not capture assistant output (avoids self-poisoning)."
    }
  }
}