@cachly-dev/mcp-server 0.9.2 → 0.9.3
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +56 -1
- package/dist/index.js +656 -2
- package/dist/index.js.map +1 -1
- package/package.json +1 -1
- package/src/index.ts +711 -2
package/CHANGELOG.md
CHANGED
|
@@ -7,7 +7,62 @@
|
|
|
7
7
|
|
|
8
8
|
---
|
|
9
9
|
|
|
10
|
-
## [0.
|
|
10
|
+
## [0.9.3] – 2026-05-06
|
|
11
|
+
|
|
12
|
+
### 🧠 10x Vision Phase 2 — Layers 3, 5 & 6
|
|
13
|
+
|
|
14
|
+
#### Layer 3 — Multi-Agent Deliberation Chamber (MADC)
|
|
15
|
+
|
|
16
|
+
New tool: **`madc_deliberate`** — When conflicting lessons exist for a topic, run deliberation between 6 specialist expert agents (InfraAgent, AuthAgent, DeployAgent, DatabaseAgent, DebugAgent, APIAgent). Each agent votes based on its CKG domain coverage. Unanimous → loser superseded. Split → contested flag. Resolutions stored as permanent CKG nodes.
|
|
17
|
+
|
|
18
|
+
Auto-trigger: `learn_from_attempts` now detects contradictions, writes a `contradicts` CKG edge, and suggests `madc_deliberate` automatically.
|
|
19
|
+
|
|
20
|
+
#### Layer 5 — Continuous Learning Stream (CLS)
|
|
21
|
+
|
|
22
|
+
Two new tools for learning **without `session_end`**:
|
|
23
|
+
|
|
24
|
+
- **`cls_ingest`** — Ingest learning signals from 3 sources: `git_commit` (commit message + files → CKG edges + lesson), `ci_outcome` (green after red → confirmed fix at 75% confidence), `ide_diagnostic` (compiler error + fix pair → instant lesson + CKG `fixes` edge).
|
|
25
|
+
|
|
26
|
+
- **`cls_install_hooks`** — Outputs a ready-to-paste git post-commit hook + GitHub Actions step. Once installed: every commit and CI run feeds the brain automatically. Zero `session_end` required.
|
|
27
|
+
|
|
28
|
+
#### Layer 6 — Federation Protocol (FedBrain)
|
|
29
|
+
|
|
30
|
+
Four new tools for AI-to-AI knowledge transfer with cryptographic provenance:
|
|
31
|
+
|
|
32
|
+
- **`fedbrain_contribute`** — Contribute a lesson to the global commons with a HMAC-signed knowledge certificate (domain fingerprint + confidence + outcome chain hash).
|
|
33
|
+
|
|
34
|
+
- **`fedbrain_search`** — Context-weighted search: your domain fingerprint (from contributed lessons + `context_hints`) weights results from brains with matching tech stacks higher. Shows Gold Standard badges (10+ confirms).
|
|
35
|
+
|
|
36
|
+
- **`fedbrain_confirm`** — Confirm that a syndicated lesson worked. Propagates back to global commons (increments `confirm_count`). Updates local CKG confidence. At 10 independent confirmations → 🏆 Gold Standard.
|
|
37
|
+
|
|
38
|
+
- **`fedbrain_status`** — Dashboard: contributed lessons, confirm history, Gold Standard count, pending propagations.
|
|
39
|
+
|
|
40
|
+
#### Summary
|
|
41
|
+
|
|
42
|
+
| Layer | Tool(s) | What it does |
|
|
43
|
+
|-------|---------|--------------|
|
|
44
|
+
| 3 (MADC) | `madc_deliberate` | Expert agent voting on contradicting beliefs |
|
|
45
|
+
| 5 (CLS) | `cls_ingest`, `cls_install_hooks` | Continuous learning without session_end |
|
|
46
|
+
| 6 (FedBrain) | `fedbrain_contribute`, `fedbrain_search`, `fedbrain_confirm`, `fedbrain_status` | Federated knowledge with crypto certificates |
|
|
47
|
+
|
|
48
|
+
Total tools: **63** (was 54 in v0.9.2)
|
|
49
|
+
|
|
50
|
+
---
|
|
51
|
+
|
|
52
|
+
## [0.9.2] – 2026-05-06
|
|
53
|
+
|
|
54
|
+
### 🧠 10x Vision Phase 1 — Layers 1, 2, 4 & 7
|
|
55
|
+
|
|
56
|
+
- **Layer 1 (CKG):** Redis-backed Causal Knowledge Graph. Typed edges: `fixes`, `requires`, `co-occurs`, `causes`, `contradicts`.
|
|
57
|
+
- **Layer 2 (BUE):** `learn_from_attempts` now writes CKG edges with Bayesian confidence `(s+1)/(t+2)`.
|
|
58
|
+
- **Layer 4 (PPE):** `brain_predict` — CKG traversal + text fallback for failure prediction.
|
|
59
|
+
- **Layer 7 (MCM):** Domain coverage map in `session_start` with confidence bars.
|
|
60
|
+
- New: `brain_search` (BM25+ over all brain data), `ckg_inspect` (BFS with confidence bars).
|
|
61
|
+
- `causal_trace` upgraded: CKG graph-first, text similarity fallback.
|
|
62
|
+
|
|
63
|
+
---
|
|
64
|
+
|
|
65
|
+
|
|
11
66
|
|
|
12
67
|
### 🌐 Knowledge Syndication — The Global AI Brain
|
|
13
68
|
|