hebbian 0.7.0 → 0.8.0

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/README.md CHANGED
@@ -2,7 +2,7 @@
2
2
  <img src="https://img.shields.io/badge/TypeScript-6.0-3178C6?style=flat-square&logo=typescript" />
3
3
  <img src="https://img.shields.io/badge/Node.js-22+-339933?style=flat-square&logo=node.js" />
4
4
  <img src="https://img.shields.io/badge/Runtime_Deps-0-brightgreen?style=flat-square" />
5
- <img src="https://img.shields.io/badge/Tests-277-blue?style=flat-square" />
5
+ <img src="https://img.shields.io/badge/Tests-317-blue?style=flat-square" />
6
6
  <img src="https://img.shields.io/badge/MIT-green?style=flat-square" />
7
7
  </p>
8
8
 
@@ -158,16 +158,56 @@ hebbian doctor # full diagnostic
158
158
 
159
159
  ---
160
160
 
161
+ ## Tool Failure Detection (v0.6.0+)
162
+
163
+ hebbian automatically learns from failed commands — no explicit correction needed:
164
+
165
+ ```bash
166
+ # During a session, a bash command fails (exit code ≠ 0)
167
+ # → hebbian digest auto-logs it as a tool-failure episode
168
+ # → evolve sees the pattern and proposes inhibitory neurons
169
+
170
+ hebbian sessions # see tool-failure episodes in the log
171
+ ```
172
+
173
+ Retry patterns (same error 3+ times) are flagged separately as `retry-pattern` episodes.
174
+
175
+ ---
176
+
177
+ ## Multi-Brain (v0.7.0+)
178
+
179
+ Per-agent brains for multi-agent setups:
180
+
181
+ ```bash
182
+ # Individual brain for each agent
183
+ hebbian grow cortex/SOME_RULE --agent cto --brain ./brain
184
+ hebbian grow cortex/OTHER_RULE --agent coo --brain ./brain
185
+
186
+ # Results in:
187
+ # brain/agents/cto/cortex/SOME_RULE/
188
+ # brain/agents/coo/cortex/OTHER_RULE/
189
+
190
+ # Shared brain (cross-cutting knowledge)
191
+ # brain/shared/cortex/...
192
+ ```
193
+
194
+ ---
195
+
161
196
  ## LLM Evolution
162
197
 
163
198
  ```bash
164
199
  GEMINI_API_KEY=... hebbian evolve --dry-run --brain ./brain
200
+
201
+ # Pruning mode (nightly cleaner — remove stale/redundant neurons)
202
+ GEMINI_API_KEY=... hebbian evolve prune --dry-run --brain ./brain
165
203
  ```
166
204
 
167
205
  The evolve engine reads the last 100 episodes + current brain state, sends it to Gemini, and proposes up to 10 mutations per cycle. Protected regions (brainstem/limbic/sensors) are blocked.
168
206
 
169
207
  Actions it can take: `grow` (new neuron), `fire` (strengthen), `signal` (dopamine/bomb), `prune` (weaken), `decay` (mark dormant).
170
208
 
209
+ **Pruning mode** uses a cleanup-focused prompt that only removes: stale neurons (30+ days inactive), high contra ratio (>0.7), redundant duplicates. Run nightly via cron.
210
+
171
211
  ---
172
212
 
173
213
  ## CLI Reference
@@ -200,6 +240,11 @@ hebbian digest [--transcript <path>]
200
240
 
201
241
  # Evolution
202
242
  GEMINI_API_KEY=... hebbian evolve [--dry-run]
243
+ GEMINI_API_KEY=... hebbian evolve prune [--dry-run] # Pruning mode (청소부)
244
+
245
+ # Multi-brain (per-agent)
246
+ hebbian grow cortex/RULE --agent cto # Routes to brain/agents/cto/
247
+ hebbian emit claude --agent coo # Emits from brain/agents/coo/
203
248
  ```
204
249
 
205
250
  ### Emit Targets
@@ -219,7 +264,7 @@ GEMINI_API_KEY=... hebbian evolve [--dry-run]
219
264
 
220
265
  | Feature | .cursorrules / CLAUDE.md | Mem0 / MemOS | hebbian |
221
266
  |---------|--------------------------|-------------|------|
222
- | Self-learning | ❌ manual | ✅ vector DB | ✅ filesystem |
267
+ | Self-learning | ❌ manual | ✅ vector DB | ✅ filesystem + tool failures |
223
268
  | Infrastructure | $0 | $$$ | **$0** |
224
269
  | Switch AI | Manual migration | Full re-setup | **`cp -r brain/`** |
225
270
  | Immutable guardrails | None | None | **brainstem + bomb** |