brain-memory 0.1.0-beta.3 → 0.1.0-beta.31
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 +339 -0
- package/README.md +400 -94
- package/SECURITY.md +61 -0
- package/assets/brand/README.md +28 -0
- package/assets/brand/icon-dark.svg +19 -0
- package/assets/brand/icon-light.svg +20 -0
- package/assets/brand/linkedin_cover.png +0 -0
- package/assets/brand/linkedin_cover_light.png +0 -0
- package/assets/brand/mark-dark.svg +19 -0
- package/assets/brand/mark-light.svg +20 -0
- package/assets/brand/png/icon-dark-1024.png +0 -0
- package/assets/brand/png/icon-dark-128.png +0 -0
- package/assets/brand/png/icon-dark-256.png +0 -0
- package/assets/brand/png/icon-dark-512.png +0 -0
- package/assets/brand/png/icon-dark-64.png +0 -0
- package/assets/brand/png/icon-light-1024.png +0 -0
- package/assets/brand/png/icon-light-128.png +0 -0
- package/assets/brand/png/icon-light-256.png +0 -0
- package/assets/brand/png/icon-light-512.png +0 -0
- package/assets/brand/png/icon-light-64.png +0 -0
- package/assets/brand/png/mark-dark-1024.png +0 -0
- package/assets/brand/png/mark-light-1024.png +0 -0
- package/assets/icon.png +0 -0
- package/assets/icon.svg +19 -0
- package/bin/brain.js +111 -0
- package/bin/cloud-sync.js +150 -0
- package/bin/forget.js +122 -0
- package/bin/install.js +229 -254
- package/bin/memorize.js +458 -0
- package/bin/pin.js +44 -0
- package/bin/recall.js +259 -0
- package/bin/reinforce.js +187 -0
- package/bin/session-start.js +298 -0
- package/bin/skill.js +67 -0
- package/bin/unpin.js +38 -0
- package/commands/brain/forget.md +245 -15
- package/commands/brain/memorize.md +76 -144
- package/commands/brain/pin.md +79 -0
- package/commands/brain/remember.md +48 -105
- package/commands/brain/skills.md +60 -0
- package/commands/brain/sleep.md +69 -28
- package/commands/brain/status.md +24 -10
- package/commands/brain/sync.md +386 -0
- package/package.json +22 -8
- package/prompts/antigravity.md +180 -0
- package/prompts/claude.md +124 -20
- package/prompts/openai.md +124 -20
- package/prompts/opencode.md +179 -0
- package/src/cloud-sync.js +534 -0
- package/src/crypto.js +130 -0
- package/src/export-import.js +200 -0
- package/src/git-sync.js +444 -0
- package/src/index-manager.js +219 -18
- package/src/installer.js +444 -0
- package/src/pinning.js +124 -0
- package/src/scorer.js +45 -11
- package/src/skills.js +208 -0
- package/src/tfidf.js +520 -0
- package/commands/brain/consolidate.md +0 -153
- package/commands/brain/explore.md +0 -79
- package/commands/brain/init.md +0 -157
- package/commands/brain/review.md +0 -130
- package/commands/brain/sunshine.md +0 -337
- package/hooks/session-end.md +0 -48
- package/hooks/session-start.md +0 -48
- package/prompts/gemini.md +0 -76
package/README.md
CHANGED
|
@@ -1,13 +1,21 @@
|
|
|
1
|
-
|
|
1
|
+
<p align="center">
|
|
2
|
+
<img src="assets/icon.png" alt="Brain Memory" width="128" height="128">
|
|
3
|
+
</p>
|
|
2
4
|
|
|
3
|
-
|
|
5
|
+
<h1 align="center">Brain Memory</h1>
|
|
6
|
+
|
|
7
|
+
<p align="center">
|
|
8
|
+
<a href="https://github.com/omelas-tech/brain/actions/workflows/ci.yml"><img src="https://github.com/omelas-tech/brain/actions/workflows/ci.yml/badge.svg" alt="CI"></a>
|
|
9
|
+
<a href="https://www.npmjs.com/package/brain-memory"><img src="https://img.shields.io/npm/v/brain-memory" alt="npm version"></a>
|
|
10
|
+
<a href="https://github.com/omelas-tech/brain/blob/main/LICENSE"><img src="https://img.shields.io/npm/l/brain-memory" alt="license"></a>
|
|
11
|
+
</p>
|
|
4
12
|
|
|
5
13
|
A hierarchical, file-system-based memory plugin for AI coding agents. Inspired by human neuroscience — memories are organized into deep nested life-domain categories, connected via associative networks, strengthened through spaced recall, and naturally decay over time.
|
|
6
14
|
|
|
7
|
-
|
|
15
|
+
One hosted **MCP connector** reaches every major agent and chat app — **Claude Code**, **OpenAI Codex CLI**, **OpenCode**, the **Claude.ai** apps, **ChatGPT**, and **Google Antigravity** (Gemini CLI's successor). Prefer a free, local-first install? The native plugin runs entirely from `~/.brain/` on Claude Code, Codex, OpenCode, and Antigravity.
|
|
8
16
|
|
|
9
17
|
```
|
|
10
|
-
|
|
18
|
+
~/.brain/
|
|
11
19
|
├── professional/
|
|
12
20
|
│ └── companies/
|
|
13
21
|
│ └── acme-corp/
|
|
@@ -36,65 +44,205 @@ Existing AI memory solutions use flat databases with tag-based retrieval. Brain
|
|
|
36
44
|
- **Context-dependent recall** — Memories encoded in a similar context to the current session are scored higher
|
|
37
45
|
- **Spaced reinforcement** — Memories recalled after longer intervals get bigger boosts, cramming produces diminishing returns
|
|
38
46
|
- **Cognitive types** — Episodic, semantic, and procedural memories each decay differently, just like in the brain
|
|
47
|
+
- **Always-present knowledge** — Pin critical conventions and preferences so they load every session, bypassing recall and never decaying — the facts your agent must never miss
|
|
48
|
+
- **Procedural skills** — Reusable how-to workflows with progressive disclosure, learned automatically from repeated experience and exportable to your agent's native skills
|
|
39
49
|
- **On-demand depth** — Subcategories are created as needed, not pre-defined
|
|
40
50
|
- **Consolidation** — Weak related memories merge into stronger combined knowledge
|
|
41
51
|
- **Zero dependencies** — Pure file I/O, no databases, no servers, no embeddings required
|
|
42
52
|
|
|
43
53
|
## Install
|
|
44
54
|
|
|
55
|
+
There are two ways to give an agent its brain. They are complementary — most people end up with both:
|
|
56
|
+
|
|
57
|
+
| Path | Reaches | Storage | Cost | Best for |
|
|
58
|
+
|------|---------|---------|------|----------|
|
|
59
|
+
| **Hosted MCP connector** | Every MCP-capable host (Claude Code, Codex CLI, OpenCode, Claude.ai apps, ChatGPT, Antigravity) | Brain Cloud (hosted) | account | One connector, everywhere — including web & mobile chat |
|
|
60
|
+
| **Local-first native plugin** | The current CLIs (Claude Code, Codex CLI, OpenCode, Antigravity) | `~/.brain/` on your machine | free | Zero-config ambient memory, no account, files you own |
|
|
61
|
+
|
|
62
|
+
### 1. Universal path — the hosted MCP connector
|
|
63
|
+
|
|
64
|
+
Brain Cloud exposes a single remote MCP server (streamable-HTTP + OAuth) at **`https://mcp.brainmemory.ai/mcp`**. Add it once per host and that host can recall, memorize, pin, and check status — no local install, no per-project setup. It works everywhere MCP does, including the chat apps that can't run a native plugin.
|
|
65
|
+
|
|
45
66
|
```bash
|
|
46
|
-
|
|
67
|
+
# Claude Code
|
|
68
|
+
claude mcp add --transport http brain https://mcp.brainmemory.ai/mcp
|
|
69
|
+
|
|
70
|
+
# OpenAI Codex CLI
|
|
71
|
+
codex mcp add brain --url https://mcp.brainmemory.ai/mcp
|
|
47
72
|
```
|
|
48
73
|
|
|
49
|
-
|
|
74
|
+
<details>
|
|
75
|
+
<summary>Per-host setup for every other connector</summary>
|
|
76
|
+
|
|
77
|
+
- **Codex CLI (config file)** — add to `~/.codex/config.toml`:
|
|
78
|
+
```toml
|
|
79
|
+
[mcp_servers.brain]
|
|
80
|
+
url = "https://mcp.brainmemory.ai/mcp"
|
|
81
|
+
```
|
|
82
|
+
- **OpenCode** — add to `opencode.json`:
|
|
83
|
+
```json
|
|
84
|
+
{
|
|
85
|
+
"mcp": {
|
|
86
|
+
"brain": { "type": "remote", "url": "https://mcp.brainmemory.ai/mcp" }
|
|
87
|
+
}
|
|
88
|
+
}
|
|
89
|
+
```
|
|
90
|
+
- **Claude.ai / ChatGPT** — **Settings → Connectors → Add custom connector**, paste `https://mcp.brainmemory.ai/mcp`, and complete the OAuth sign-in. (ChatGPT requires a paid plan or Developer Mode.)
|
|
91
|
+
- **Google Antigravity** — add to `mcp_config.json` under the `serverUrl` key. *(Legacy Gemini CLI used the `httpUrl` key in `settings.json`.)*
|
|
50
92
|
|
|
51
|
-
|
|
93
|
+
</details>
|
|
94
|
+
|
|
95
|
+
The first time a host connects you'll complete an OAuth sign-in with the Google account that holds your cloud brain; the connector then syncs to every client on that account. See **[Use Brain in Claude apps](https://brainmemory.ai/docs/getting-started/claude-apps)** for the full walkthrough.
|
|
96
|
+
|
|
97
|
+
### 2. Local-first path — the native plugin
|
|
52
98
|
|
|
53
99
|
```bash
|
|
54
|
-
|
|
55
|
-
|
|
56
|
-
npx brain-memory --codex --global # OpenAI Codex CLI, global
|
|
57
|
-
npx brain-memory --all --global # All runtimes, global
|
|
100
|
+
npm install -g brain-memory@beta
|
|
101
|
+
brain
|
|
58
102
|
```
|
|
59
103
|
|
|
104
|
+
The first command installs the package globally — this gives you both the interactive setup wizard and the `brain` CLI (`brain recall`, `brain memorize`, `brain reinforce`, …) that agents rely on for deterministic scoring.
|
|
105
|
+
|
|
106
|
+
The second command runs the setup wizard, which asks which runtime(s) to configure and whether to install globally or for the current project. Each runtime gets its instructions file plus the brain commands/skills:
|
|
107
|
+
|
|
108
|
+
| Runtime | Instructions file | Commands land at |
|
|
109
|
+
|---------|-------------------|------------------|
|
|
110
|
+
| **Claude Code** | `CLAUDE.md` | `~/.claude/commands/brain/` (plugin slash commands) |
|
|
111
|
+
| **OpenAI Codex CLI** | `AGENTS.md` (`~/.codex/`) | `~/.agents/skills/<name>/SKILL.md` |
|
|
112
|
+
| **OpenCode** | `AGENTS.md` | `~/.config/opencode/commands/` |
|
|
113
|
+
| **Google Antigravity** | `GEMINI.md` | `~/.gemini/skills/` *(experimental — verify paths against a live install)* |
|
|
114
|
+
|
|
115
|
+
#### Non-interactive
|
|
116
|
+
|
|
117
|
+
```bash
|
|
118
|
+
brain --claude --global # Claude Code, global
|
|
119
|
+
brain --codex --global # OpenAI Codex CLI, global
|
|
120
|
+
brain --opencode --global # OpenCode, global
|
|
121
|
+
brain --antigravity --global # Google Antigravity, global (experimental)
|
|
122
|
+
brain --all --global # All runtimes, global
|
|
123
|
+
```
|
|
124
|
+
|
|
125
|
+
### Update
|
|
126
|
+
|
|
127
|
+
```bash
|
|
128
|
+
npm install -g brain-memory@beta
|
|
129
|
+
brain update
|
|
130
|
+
```
|
|
131
|
+
|
|
132
|
+
The first command updates the package and the `brain` CLI. The second command refreshes the slash command prompts for your installed runtimes. Target specific runtimes with `--claude`, `--codex`, `--opencode`, `--antigravity`, or `--all`.
|
|
133
|
+
|
|
134
|
+
> **Upgrading from an older version?** The separate binaries (`brain-recall`, `brain-memorize`, `brain-reinforce`, `brain-cloud`, `brain-memory`) were unified into a single `brain` dispatcher — use `brain recall`, `brain memorize`, `brain reinforce`, `brain cloud <…>` instead. Run `brain update` to refresh your runtime prompts to the new command surface.
|
|
135
|
+
|
|
136
|
+
> **Why not `npx`?** `npx` runs the setup wizard in a temporary directory that is discarded after execution. The `brain` CLI (`brain recall`, `brain memorize`, `brain reinforce`) won't be available in your PATH, which means agents will fall back to less reliable manual file operations. Always use `npm install -g` to ensure everything works correctly.
|
|
137
|
+
|
|
138
|
+
### Uninstall
|
|
139
|
+
|
|
140
|
+
```bash
|
|
141
|
+
brain uninstall
|
|
142
|
+
npm uninstall -g brain-memory
|
|
143
|
+
```
|
|
144
|
+
|
|
145
|
+
The first command removes slash commands and prompt sections from your agent configs. The second removes the package and CLI tools. Your `~/.brain/` directory (memories) is preserved by default. Add `--delete-data` to the uninstall command to remove it too. Use `--yes` to skip confirmation prompts.
|
|
146
|
+
|
|
60
147
|
### Manual Install
|
|
61
148
|
|
|
62
|
-
|
|
149
|
+
If you'd rather wire up the native plugin by hand, copy the brain commands into your agent's folder, then append the matching prompt file to its instructions file.
|
|
63
150
|
|
|
64
151
|
```bash
|
|
65
|
-
# Claude Code
|
|
152
|
+
# Claude Code — flat slash commands
|
|
66
153
|
cp -r commands/brain/ ~/.claude/commands/brain/
|
|
67
154
|
|
|
68
|
-
#
|
|
69
|
-
|
|
155
|
+
# OpenCode — flat slash commands
|
|
156
|
+
mkdir -p ~/.config/opencode/commands/brain
|
|
157
|
+
cp -r commands/brain/. ~/.config/opencode/commands/brain/
|
|
158
|
+
|
|
159
|
+
# OpenAI Codex CLI — each command becomes a skill under the cross-tool skills dir
|
|
160
|
+
for f in commands/brain/*.md; do
|
|
161
|
+
name=$(basename "$f" .md)
|
|
162
|
+
mkdir -p ~/.agents/skills/brain-"$name"
|
|
163
|
+
cp "$f" ~/.agents/skills/brain-"$name"/SKILL.md
|
|
164
|
+
done
|
|
70
165
|
|
|
71
|
-
#
|
|
166
|
+
# Google Antigravity (experimental — verify paths) — skills under ~/.gemini/skills/
|
|
72
167
|
for f in commands/brain/*.md; do
|
|
73
168
|
name=$(basename "$f" .md)
|
|
74
|
-
mkdir -p ~/.
|
|
75
|
-
cp "$f" ~/.
|
|
169
|
+
mkdir -p ~/.gemini/skills/brain-"$name"
|
|
170
|
+
cp "$f" ~/.gemini/skills/brain-"$name"/SKILL.md
|
|
76
171
|
done
|
|
77
172
|
```
|
|
78
173
|
|
|
79
174
|
Then append the contents of the corresponding prompt file to your agent's instructions file:
|
|
80
175
|
- `prompts/claude.md` → `CLAUDE.md`
|
|
81
|
-
- `prompts/
|
|
82
|
-
- `prompts/
|
|
176
|
+
- `prompts/openai.md` → `AGENTS.md` (Codex, in `~/.codex/`)
|
|
177
|
+
- `prompts/opencode.md` → `AGENTS.md` (OpenCode)
|
|
178
|
+
- `prompts/antigravity.md` → `GEMINI.md` (experimental)
|
|
179
|
+
|
|
180
|
+
> **`AGENTS.md` collision (local scope):** Codex and OpenCode both read a project-local `AGENTS.md`. If you install both for the *same* project, point only one at the local file (or install one globally) so their brain sections don't conflict. Globally they're separate (`~/.codex/AGENTS.md` vs OpenCode's own config dir).
|
|
83
181
|
|
|
84
182
|
## Commands
|
|
85
183
|
|
|
184
|
+
The everyday loop is **ambient** — `remember` and `memorize` run automatically at session start/end, so you rarely type a command. The surface is a small core plus a background maintenance job:
|
|
185
|
+
|
|
86
186
|
| Command | Description |
|
|
87
187
|
|---------|-------------|
|
|
88
|
-
| `/brain:init` | Initialize `.brain/` directory structure with default categories |
|
|
89
|
-
| `/brain:memorize [topic]` | Store memories from current session context |
|
|
90
188
|
| `/brain:remember [query]` | Recall relevant memories with spreading activation and context matching |
|
|
91
|
-
| `/brain:
|
|
92
|
-
| `/brain:explore [category]` | Browse the brain hierarchy with visual tree view |
|
|
93
|
-
| `/brain:consolidate [scope]` | Merge related weak memories into stronger combined ones |
|
|
94
|
-
| `/brain:forget [target]` | Decay, archive, or remove memories |
|
|
95
|
-
| `/brain:sunshine [target]` | Deep forensic erasure — trace and remove all references |
|
|
96
|
-
| `/brain:sleep [scope]` | Full maintenance cycle — 9 neuroscience-inspired phases |
|
|
189
|
+
| `/brain:memorize [topic] [--sync]` | Store memories from current session context (add `--sync` to auto-push) |
|
|
97
190
|
| `/brain:status` | Dashboard with brain health metrics and recommendations |
|
|
191
|
+
| `/brain:pin [id\|query]` | Pin a memory to the always-present tier — loads every session, never decays. Toggles: also unpins (`--off`) |
|
|
192
|
+
| `/brain:forget [target]` | Decay, archive, or remove memories. `--deep` performs forensic erasure — traces and removes every reference |
|
|
193
|
+
| `/brain:sync [subcommand]` | Sync via Brain Cloud, Git remote, or export/import (auto-initializes the brain on first run) |
|
|
194
|
+
| `/brain:skills [list\|show\|add\|use\|remove\|export]` | Manage procedural skills — reusable how-to workflows with progressive disclosure |
|
|
195
|
+
| `/brain:sleep [scope]` | Full maintenance cycle — 9 neuroscience-inspired phases (replay, consolidation, review reinforcement, pruning, dreaming, …). Usually runs automatically/in the background |
|
|
196
|
+
|
|
197
|
+
## Session Lifecycle
|
|
198
|
+
|
|
199
|
+
Brain Memory works automatically in the background — no commands needed for basic context awareness.
|
|
200
|
+
|
|
201
|
+
### Session Start
|
|
202
|
+
|
|
203
|
+
When a session begins and `~/.brain/` exists, the agent makes a single `brain session-start` call. This aggregator returns one deterministic, **token-budget-bounded** payload so the brain can never bloat the context window (the budget lives in `~/.brain/config.json`):
|
|
204
|
+
|
|
205
|
+
1. **Pinned memories** — always-present conventions and preferences, injected verbatim regardless of recall score
|
|
206
|
+
2. **Skills index** — the name + description of each available procedural skill (~100 tokens each; full instructions are loaded only when a task matches)
|
|
207
|
+
3. **Context recall** — the top memories relevant to the current project
|
|
208
|
+
4. **Review queue + low-confidence alerts**, then a brief status line:
|
|
209
|
+
|
|
210
|
+
```
|
|
211
|
+
◉ Brain active — 42 memories loaded (8 in current project context)
|
|
212
|
+
📋 3 memories due for review — reinforced automatically during /brain:sleep
|
|
213
|
+
```
|
|
214
|
+
|
|
215
|
+
The agent treats pinned facts as active constraints, notes which skills exist, silently internalizes relevant memories, and references them naturally during the session — no information dump.
|
|
216
|
+
|
|
217
|
+
### Ambient Session Tracking
|
|
218
|
+
|
|
219
|
+
Throughout the session, the agent maintains a running mental log of notable events — decisions made, things learned, insights realized, significant experiences. This is purely internal awareness with no file writes. It ensures that early-session events aren't forgotten by the time the session ends.
|
|
220
|
+
|
|
221
|
+
### Periodic Memory Checkpoint
|
|
222
|
+
|
|
223
|
+
Every ~10 substantive interactions (file edits, architecture decisions, debugging breakthroughs), the agent evaluates whether memorizable content has accumulated. If so, it appends a brief reminder to its next response:
|
|
224
|
+
|
|
225
|
+
```
|
|
226
|
+
◉ Notable decisions and learnings this session — /brain:memorize when ready
|
|
227
|
+
```
|
|
228
|
+
|
|
229
|
+
This never interrupts your flow — it's a one-liner appended to an existing response, at most once per ~10 interactions. The counter resets after you run `/brain:memorize`.
|
|
230
|
+
|
|
231
|
+
### Session End
|
|
232
|
+
|
|
233
|
+
When a session ends, the agent performs two steps in order:
|
|
234
|
+
|
|
235
|
+
1. **Saves session context** to `~/.brain/contexts.json` — always, even for trivial sessions. This includes a `notable_unsaved` field listing items you didn't memorize, so future sessions can reference what happened.
|
|
236
|
+
|
|
237
|
+
2. **Suggests memorization** if the session contained meaningful decisions, learnings, or insights:
|
|
238
|
+
|
|
239
|
+
```
|
|
240
|
+
💡 This session contained notable decisions and learnings.
|
|
241
|
+
Would you like to store them as brain memories?
|
|
242
|
+
Run /brain:memorize to capture them before this context is lost.
|
|
243
|
+
```
|
|
244
|
+
|
|
245
|
+
The agent never auto-memorizes without user consent. Context is saved proactively — the agent doesn't wait for an explicit goodbye signal.
|
|
98
246
|
|
|
99
247
|
## How It Works
|
|
100
248
|
|
|
@@ -111,13 +259,13 @@ Create → Store → Decay → Recall → Reinforce → Review → Sleep → Arc
|
|
|
111
259
|
3. **Decay** — Memories naturally weaken over time: `effective_strength = base_strength * (decay_rate ^ days_since_access)`
|
|
112
260
|
4. **Recall** — `/brain:remember` searches, scores with spreading activation and context matching, and returns the best memories. Archived memories are searched as a fallback.
|
|
113
261
|
5. **Reinforce** — Each recall applies spaced reinforcement (longer gaps = bigger boosts) and improves the memory's decay resistance
|
|
114
|
-
6. **Review** —
|
|
115
|
-
7. **Sleep** — `/brain:sleep` performs a 9-phase maintenance cycle inspired by real neuroscience
|
|
262
|
+
6. **Review** — SM-2 spaced repetition surfaces memories at optimal intervals for long-term retention; reinforcement runs automatically during the sleep cycle
|
|
263
|
+
7. **Sleep** — `/brain:sleep` performs a 9-phase maintenance cycle inspired by real neuroscience (replay, consolidation, review reinforcement, pruning, dreaming, …), usually run automatically/in the background
|
|
116
264
|
8. **Archive** — Fully decayed memories move to `_archived/` (recoverable and searchable) or can be permanently deleted
|
|
117
265
|
|
|
118
266
|
### Neuroscience Foundations
|
|
119
267
|
|
|
120
|
-
Brain Memory
|
|
268
|
+
Brain Memory is grounded in peer-reviewed neuroscience research. Here's how each mechanism maps to the brain:
|
|
121
269
|
|
|
122
270
|
| Brain Mechanism | Implementation |
|
|
123
271
|
|-----------------|---------------|
|
|
@@ -158,6 +306,38 @@ Each memory is also classified by how the brain processes it:
|
|
|
158
306
|
|
|
159
307
|
During sleep, frequently-recalled episodic memories are **crystallized** into semantic memories — the specific event fades but the lesson persists. This mirrors how humans extract general principles from repeated experiences.
|
|
160
308
|
|
|
309
|
+
### Always-Present Memories (Pinning)
|
|
310
|
+
|
|
311
|
+
Recall is probabilistic — a stored preference only applies if scoring happens to surface it. For the facts an agent must **never** miss (coding conventions, standing decisions, hard constraints), that's not good enough. Pinning fixes it:
|
|
312
|
+
|
|
313
|
+
- **`pinned`** — the memory is injected at **every** session start regardless of recall score, and is **decay-exempt** (it never fades). Scope it `global` (loads everywhere) or `project:<name>` (loads only in that project).
|
|
314
|
+
- **`stable`** — an independent flag that exempts a memory from decay and pruning **without** forcing it to always load — for timeless facts you recall on demand but don't want to fade.
|
|
315
|
+
|
|
316
|
+
```bash
|
|
317
|
+
brain pin <id> --scope global --priority 1 # or: /brain:pin always use tabs
|
|
318
|
+
```
|
|
319
|
+
|
|
320
|
+
Pinned and stable memories are skipped by sleep-cycle homeostasis and pruning, and `/brain:memorize` proactively proposes pinning durable conventions. The always-present set is budget-capped (`pin_budget_tokens`) so it can't crowd out the context window. This maps cleanly to the **semantic memory** type in the [CoALA](https://arxiv.org/abs/2309.02427) agent-memory model — knowledge that is always in context — decomposed into two orthogonal properties (always-loaded vs. non-decaying).
|
|
321
|
+
|
|
322
|
+
### Procedural Skills
|
|
323
|
+
|
|
324
|
+
Procedural memory is *how* to do things — reusable, step-by-step workflows stored as `~/.brain/_skills/<name>/SKILL.md`. To keep them from flooding the context window, skills use **three-level progressive disclosure**:
|
|
325
|
+
|
|
326
|
+
| Level | When | What loads |
|
|
327
|
+
|-------|------|------------|
|
|
328
|
+
| **L0** | Every session start | Only each skill's name + description (~100 tokens) |
|
|
329
|
+
| **L1** | A task matches a skill | The full `SKILL.md` step-by-step instructions |
|
|
330
|
+
| **L2** | A step needs them | Referenced `resources/` (templates, scripts) |
|
|
331
|
+
|
|
332
|
+
```bash
|
|
333
|
+
brain skill list # advertised skills (L0)
|
|
334
|
+
brain skill show structured-code-review # full instructions (L1)
|
|
335
|
+
brain skill use structured-code-review # record outcome (--failed weakens it)
|
|
336
|
+
brain skill export structured-code-review --target claude # → native .claude/skills/
|
|
337
|
+
```
|
|
338
|
+
|
|
339
|
+
Skills strengthen on successful `use` and weaken on `--failed` — one that fails too often demotes itself out of the advertised L0 index. And they aren't only hand-authored: during sleep, Brain **crystallizes** recurring task-solving patterns from repeated experience into new skills (user-confirmed). Finally, `brain skill export` emits a skill in your agent's **native** format so it becomes directly executable.
|
|
340
|
+
|
|
161
341
|
### Memory File Format
|
|
162
342
|
|
|
163
343
|
Each memory is a Markdown file with YAML frontmatter:
|
|
@@ -226,7 +406,15 @@ score = 0.38 * relevance
|
|
|
226
406
|
- **context_match** (0.14) — How similar the encoding context is to the current session
|
|
227
407
|
- **salience** (0.08) — Emotional/motivational significance
|
|
228
408
|
|
|
229
|
-
|
|
409
|
+
Relevance is **calibrated in absolute terms**: BM25 scores are scaled by
|
|
410
|
+
IDF-weighted *query coverage*, so a memory matching one term of a four-term query
|
|
411
|
+
cannot claim relevance 1.0 just for being the best available match. Explicit-query
|
|
412
|
+
recall also applies a **relevance floor** — memories that are neither
|
|
413
|
+
query-relevant nor activated by a relevant associate are excluded rather than
|
|
414
|
+
padding the results on strength alone. A query about things the brain doesn't
|
|
415
|
+
know returns few results with honestly low scores (or none), instead of
|
|
416
|
+
confident-looking noise. Context-mode recall (session start) is exempt, where
|
|
417
|
+
strength-ranked topical padding is the intended behavior.
|
|
230
418
|
|
|
231
419
|
The agent then decides the response strategy:
|
|
232
420
|
- **Single strong match** (top score > 0.7) → Return the full memory
|
|
@@ -236,7 +424,7 @@ The agent then decides the response strategy:
|
|
|
236
424
|
|
|
237
425
|
### Associative Network
|
|
238
426
|
|
|
239
|
-
Memories are connected via weighted edges in
|
|
427
|
+
Memories are connected via weighted edges in `~/.brain/associations.json`:
|
|
240
428
|
|
|
241
429
|
```json
|
|
242
430
|
{
|
|
@@ -261,7 +449,7 @@ Memories are connected via weighted edges in `.brain/associations.json`:
|
|
|
261
449
|
|
|
262
450
|
### Spaced Reinforcement
|
|
263
451
|
|
|
264
|
-
|
|
452
|
+
Spaced reinforcement rewards optimal recall timing:
|
|
265
453
|
|
|
266
454
|
```
|
|
267
455
|
spacingMultiplier = min(3.0, 1.0 + log2(1 + daysSinceLastAccess))
|
|
@@ -320,7 +508,7 @@ Each expertise area gets an `_expertise.md` profile documenting what you know we
|
|
|
320
508
|
|
|
321
509
|
### Spaced Repetition Review
|
|
322
510
|
|
|
323
|
-
|
|
511
|
+
Brain Memory implements the SM-2 algorithm to surface memories at optimal intervals for long-term retention. The review queue is generated and reinforced during `/brain:sleep`, and tracks:
|
|
324
512
|
|
|
325
513
|
- **Interval** — Time until next review (grows exponentially with successful recalls)
|
|
326
514
|
- **Ease factor** — How easily the memory is recalled (adjusts based on recall quality 1-5)
|
|
@@ -328,24 +516,78 @@ Each expertise area gets an `_expertise.md` profile documenting what you know we
|
|
|
328
516
|
|
|
329
517
|
Failed recalls reset the interval to 1 day. Successful recalls extend the interval by the ease factor. This ensures you spend time on memories that need reinforcement, not ones you already know well.
|
|
330
518
|
|
|
331
|
-
###
|
|
519
|
+
### Cross-Agent Memory Sharing
|
|
520
|
+
|
|
521
|
+
`~/.brain/` is a single global directory in the user's home folder. All memories are shared across every project and every supported agent automatically. A decision stored by Claude Code in one project is immediately available to OpenAI Codex CLI, OpenCode, or Google Antigravity in any other project — no configuration, no export, no per-project setup. The format is agent-agnostic: plain Markdown files with YAML frontmatter, readable by any tool.
|
|
522
|
+
|
|
523
|
+
Brain Memory is **model-agnostic** as well as agent-portable: because memory is plain files rather than embeddings welded to a particular model, the LLM underneath your agent can be anything — GPT, Claude, Gemini, or any model routed through a gateway — and your memory is unaffected. **Switch your model or switch your agent, and you keep remembering and pick up exactly where you left off.**
|
|
524
|
+
|
|
525
|
+
To share memories across different machines, use `/brain:sync` (see below).
|
|
332
526
|
|
|
333
|
-
|
|
334
|
-
- Creates `associations.json` from existing `related` fields
|
|
335
|
-
- Backfills `recall_history` from `access_count`
|
|
336
|
-
- Sets sensible defaults for new fields (`cognitive_type: semantic`, `salience: 0.5`, `confidence: 0.8`)
|
|
337
|
-
- Bumps `index.json` version to 2
|
|
527
|
+
### Portable Sync — your memory, no lock-in
|
|
338
528
|
|
|
339
|
-
|
|
529
|
+
**Your brain is plain files in a folder — sync it however you already sync files.** No account or cloud provider is required.
|
|
530
|
+
|
|
531
|
+
**Any synced folder** — Point `BRAIN_DIR` at a folder your existing tools already sync, and you're done:
|
|
532
|
+
|
|
533
|
+
```bash
|
|
534
|
+
export BRAIN_DIR="$HOME/Google Drive/brain" # or Dropbox, iCloud Drive, OneDrive, Syncthing…
|
|
535
|
+
```
|
|
536
|
+
|
|
537
|
+
**Git remote** — `/brain:sync` push/pull `~/.brain/` to any private Git repository (GitHub, GitLab, Codeberg, or self-hosted), using your existing Git/SSH auth.
|
|
538
|
+
|
|
539
|
+
**Export/Import** — Pack the entire brain into a single encrypted file for manual transfer via USB, email, or any file-sharing service.
|
|
540
|
+
|
|
541
|
+
**Brain Cloud** (optional) — A hosted, zero-config hub if you'd rather not run your own. Entirely optional — everything above works without it.
|
|
542
|
+
|
|
543
|
+
**Key features:**
|
|
544
|
+
- **Manual push/pull** — No background watchers, no auto-sync. You control when data moves.
|
|
545
|
+
- **Optional encryption** — AES-256-GCM encryption with a user-provided passphrase. When enabled, all files are encrypted before committing or exporting.
|
|
546
|
+
- **Merge mode** — Import supports merge mode (only import newer files) or overwrite mode.
|
|
547
|
+
- **Zero dependencies** — Uses Node.js built-in `crypto` and the system `git` binary.
|
|
548
|
+
|
|
549
|
+
**Setup:**
|
|
550
|
+
1. Create a private Git repo (e.g., `gh repo create brain-data --private`)
|
|
551
|
+
2. Run `/brain:sync setup git@github.com:you/brain-data.git`
|
|
552
|
+
3. Use `/brain:sync push` and `/brain:sync pull` to keep memories in sync
|
|
553
|
+
|
|
554
|
+
For one-off transfers, use `/brain:sync export` and `/brain:sync import <path>`.
|
|
555
|
+
|
|
556
|
+
Sync state is stored locally in `~/.brain/.sync/` and is never pushed to the remote.
|
|
557
|
+
|
|
558
|
+
### Brain in Claude apps (web & mobile) — beta
|
|
559
|
+
|
|
560
|
+
Beyond the CLI, you can use your brain from the **Claude apps** (web, desktop, and your
|
|
561
|
+
**phone**) via the **Brain connector** — a remote MCP server hosted by [Brain Cloud](https://app.brainmemory.ai).
|
|
562
|
+
Add it once on claude.ai (**Settings → Connectors → Add custom connector → `https://mcp.brainmemory.ai/mcp`**),
|
|
563
|
+
sign in with the Google account holding your cloud brain, and it syncs to every Claude client on
|
|
564
|
+
your account. Read **and write**: recall + status, plus memorize / pin / unpin (writes prompt for
|
|
565
|
+
confirmation) — so you can capture a memory on your phone and recall it on your laptop.
|
|
566
|
+
|
|
567
|
+
→ Full guide: **[Use Brain in Claude apps](https://brainmemory.ai/docs/getting-started/claude-apps)**
|
|
568
|
+
|
|
569
|
+
**Your data, protected.** Brains synced to Brain Cloud are **encrypted at rest**
|
|
570
|
+
(AES-256-GCM, per-user key); the connector holds working copies in **RAM only**
|
|
571
|
+
(never on disk); all traffic is **TLS**; sessions use **rotating, revocable tokens
|
|
572
|
+
with reuse detection** (log out one device or all); and every request is isolated
|
|
573
|
+
to your account. Prefer the server never sees plaintext? Stay **local-only** (the
|
|
574
|
+
default) or use Git/export sync with a passphrase. See **[SECURITY.md](SECURITY.md)**.
|
|
340
575
|
|
|
341
576
|
## File Structure
|
|
342
577
|
|
|
343
578
|
```
|
|
344
|
-
|
|
579
|
+
~/.brain/
|
|
345
580
|
├── index.json # Memory inventory — fast lookup for all memories
|
|
581
|
+
├── config.json # Working-memory token budgets (session-start injection)
|
|
346
582
|
├── associations.json # Weighted associative network between memories
|
|
347
583
|
├── contexts.json # Session context snapshots for context-dependent recall
|
|
348
584
|
├── review-queue.json # Spaced repetition scheduling
|
|
585
|
+
├── pinned.json # Always-present tier manifest (pinned memory IDs + scope)
|
|
586
|
+
├── skills-index.json # Advertised procedural skills (L0 — name + description)
|
|
587
|
+
├── _skills/ # Procedural skills
|
|
588
|
+
│ └── <skill-name>/
|
|
589
|
+
│ ├── SKILL.md # Advertised description + step-by-step instructions
|
|
590
|
+
│ └── resources/ # Optional templates/scripts (loaded only at execution)
|
|
349
591
|
├── professional/ # Work, career, technical skills
|
|
350
592
|
│ ├── _meta.json # Category metadata and stats
|
|
351
593
|
│ ├── _expertise.md # Generated expertise profile
|
|
@@ -368,16 +610,19 @@ No data is lost. All new scoring gracefully falls back when newer fields are abs
|
|
|
368
610
|
│ └── events/
|
|
369
611
|
├── _consolidated/ # Merged memories from consolidation
|
|
370
612
|
│ └── _meta.json
|
|
371
|
-
|
|
372
|
-
|
|
373
|
-
|
|
613
|
+
├── _archived/ # Decayed memories (recoverable + searchable)
|
|
614
|
+
│ ├── _meta.json
|
|
615
|
+
│ └── index.json # Searchable archive index
|
|
616
|
+
└── .sync/ # Sync state (local only, never pushed)
|
|
617
|
+
├── config.json # Remote URL, encryption flag
|
|
618
|
+
└── repo/ # Hidden git repo for sync
|
|
374
619
|
```
|
|
375
620
|
|
|
376
621
|
Subdirectories are created **on demand** — the agent decides placement depth based on how specific the memory is. A generic career thought lands in `professional/`, but a specific deployment incident goes to `professional/companies/acme/projects/alpha/`.
|
|
377
622
|
|
|
378
623
|
## Configuration
|
|
379
624
|
|
|
380
|
-
Brain configuration lives in
|
|
625
|
+
Brain configuration lives in `~/.brain/index.json` under the `config` key:
|
|
381
626
|
|
|
382
627
|
```json
|
|
383
628
|
{
|
|
@@ -413,53 +658,114 @@ Brain configuration lives in `.brain/index.json` under the `config` key:
|
|
|
413
658
|
| `association_config.spreading_activation_depth` | 2 | Maximum hops for spreading activation traversal |
|
|
414
659
|
| `association_config.spreading_activation_decay` | 0.5 | Decay factor per hop during spreading activation |
|
|
415
660
|
|
|
416
|
-
|
|
417
|
-
|
|
418
|
-
|
|
419
|
-
|
|
420
|
-
|
|
421
|
-
|
|
422
|
-
|
|
423
|
-
|
|
424
|
-
|
|
425
|
-
|
|
426
|
-
|
|
427
|
-
|
|
428
|
-
|
|
429
|
-
|
|
430
|
-
|
|
431
|
-
|
|
432
|
-
|
|
433
|
-
|
|
434
|
-
|
|
435
|
-
|
|
436
|
-
|
|
437
|
-
|
|
438
|
-
|
|
439
|
-
|
|
440
|
-
|
|
441
|
-
|
|
442
|
-
|
|
443
|
-
|
|
444
|
-
|
|
445
|
-
|
|
446
|
-
|
|
447
|
-
|
|
448
|
-
|
|
449
|
-
|
|
450
|
-
|
|
451
|
-
|
|
452
|
-
|
|
453
|
-
-
|
|
454
|
-
|
|
455
|
-
-
|
|
456
|
-
|
|
457
|
-
|
|
458
|
-
-
|
|
459
|
-
|
|
460
|
-
-
|
|
461
|
-
-
|
|
462
|
-
-
|
|
661
|
+
### Working-Memory Budget
|
|
662
|
+
|
|
663
|
+
A separate `~/.brain/config.json` (created lazily with safe defaults) caps how much the brain injects into the context window at session start, so the always-present tier can never crowd out your actual work:
|
|
664
|
+
|
|
665
|
+
```json
|
|
666
|
+
{
|
|
667
|
+
"working_memory_budget_tokens": 3000,
|
|
668
|
+
"pin_budget_tokens": 1500,
|
|
669
|
+
"skills_index_budget_tokens": 800,
|
|
670
|
+
"recall_budget_tokens": 700
|
|
671
|
+
}
|
|
672
|
+
```
|
|
673
|
+
|
|
674
|
+
| Setting | Default | Description |
|
|
675
|
+
|---------|---------|-------------|
|
|
676
|
+
| `working_memory_budget_tokens` | 3000 | Total token ceiling for the whole `brain session-start` payload |
|
|
677
|
+
| `pin_budget_tokens` | 1500 | Sub-budget for pinned, always-present memories |
|
|
678
|
+
| `skills_index_budget_tokens` | 800 | Sub-budget for the advertised skills index (L0) |
|
|
679
|
+
| `recall_budget_tokens` | 700 | Sub-budget for context-relevant recalled memories |
|
|
680
|
+
|
|
681
|
+
Token counts use a dependency-free heuristic stored on each memory at write time. When pins exceed their budget they're selected by `--priority` then strength, and the overflow is reported rather than silently dropped.
|
|
682
|
+
|
|
683
|
+
## Benchmark
|
|
684
|
+
|
|
685
|
+
Brain Memory ships with a controlled benchmark suite grounded in the 2025-2026 SOTA in long-term-memory evaluation. The system itself is a direct implementation of the [**CoALA**](https://arxiv.org/abs/2309.02427) agent-memory model (Sumers et al., 2023) — Pinned Tier maps to CoALA's *semantic* memory, procedural skills to *procedural* memory.
|
|
686
|
+
|
|
687
|
+
**Six scenarios**, each describable in one sentence:
|
|
688
|
+
|
|
689
|
+
| Id | Pitch | Tests |
|
|
690
|
+
|---|---|---|
|
|
691
|
+
| **A** *Noisy Project Folder* | "Your brain has 200 memories from 6 projects — does it find the 3 relevant ones?" | Retrieval under distractors (LongMemEval-S analog) |
|
|
692
|
+
| **B** *Three Sessions, One Decision* | "Postgres Monday, gRPC rewrite Wednesday, new resource Friday — still Postgres?" | Multi-session continuity + Pinned Tier ablation |
|
|
693
|
+
| **C** *The Contradiction Test* | "Tabs, then spaces, then tabs again — which version wins?" | Decay-weighted recency + contradiction handling |
|
|
694
|
+
| **D** *Skill Progressive Disclosure* | "Five skills indexed, one needed — does brain load just the one?" | CoALA Phase-2 L0/L1/L2 token efficiency |
|
|
695
|
+
| **E** *Continual Coding* | "Five bugs in order — does bug 5 finish faster than bug 1?" | Forward transfer; agent writes memories between tasks |
|
|
696
|
+
| **F** *Abstention* | "No deployment target in memory — does the agent ask or invent?" | Confabulation resistance |
|
|
697
|
+
|
|
698
|
+
**Methodology highlights** — agent under test is **DeepSeek V4 Pro** (single-shot), graded by a **cross-family judge panel** (Gemini + Gemma-4 + Qwen-3.5, majority vote — no judge shares the agent's family, [Preference Leakage 2502.01534](https://arxiv.org/abs/2502.01534), [PoLL 2404.18796](https://arxiv.org/abs/2404.18796)); deterministic distractor haystacks up to 1000 memories; real `brain session-start` / `brain recall` integration; tokens-per-successful-task as the headline efficiency metric; every scenario anchored by a no-memory floor and an oracle ceiling.
|
|
699
|
+
|
|
700
|
+
> Results (DeepSeek V4 Pro, 3 runs/arm, June 2026). **Scenario A — retrieval under 1000 distractors:** `brain-full` is the only retriever whose memories let the model succeed — it passes **100%** while both **BM25 (0%)** and a **vector store (0%)** fail to surface the oracle memories — and it does so at the **leanest tokens-per-success (3,199)** of any passing arm. **Scenario D — skills:** loading only the relevant skill passes 100% at **1,580 tok/success, ~31% leaner** than dumping all skill bodies. Scenarios B (continuity) shows brain efficient-and-correct; C (contradiction) and F (abstention) are honest **nulls** — the base model handles them without memory. At n=3 the token gaps are directional, not yet significant; the pass-rate gradient is the result. Full tables: [brainmemory.ai/docs/benchmarks/results](https://brainmemory.ai/docs/benchmarks/results).
|
|
701
|
+
|
|
702
|
+
**Retrieval calibration probe (in-vivo, July 2026).** Separate from the controlled
|
|
703
|
+
suite: a 20-query retrieval probe on a real working brain (177 memories, 1,245
|
|
704
|
+
association edges; tag-derived topical queries with known targets) measured the
|
|
705
|
+
scoring-calibration fix shipped after beta.30. Before: hit@1 90%, MRR 0.950 — and a
|
|
706
|
+
four-term nonsense control query still returned ten results, the top scoring a
|
|
707
|
+
confident-looking 0.679. After query-coverage scaling and the relevance floor:
|
|
708
|
+
**hit@1 100% / hit@3 100% / MRR 1.000**, the nonsense control returns a single result
|
|
709
|
+
with honestly low relevance (0.213), and recall latency stays at ~46 ms (p50).
|
|
710
|
+
*Caveats: n=20, single brain, tag-derived queries favor the index's tag field, not
|
|
711
|
+
judge-graded — this is a calibration probe, not the controlled benchmark above.*
|
|
712
|
+
Method and detail: [brainmemory.ai/docs/benchmarks/results](https://brainmemory.ai/docs/benchmarks/results).
|
|
713
|
+
|
|
714
|
+
Run the benchmarks yourself:
|
|
715
|
+
|
|
716
|
+
```bash
|
|
717
|
+
cd benchmark
|
|
718
|
+
cp .env.example .env # Add your API keys
|
|
719
|
+
npm test # Unit tests
|
|
720
|
+
node harness/runner.js # Full benchmark (cloud APIs)
|
|
721
|
+
```
|
|
722
|
+
|
|
723
|
+
Live methodology and per-scenario detail: [brainmemory.ai/docs/benchmarks](https://brainmemory.ai/docs/benchmarks) · Source: [`benchmark/`](benchmark/)
|
|
724
|
+
|
|
725
|
+
## Contributing
|
|
726
|
+
|
|
727
|
+
Contributions are welcome! See [CONTRIBUTING.md](CONTRIBUTING.md) for development setup, project structure, and how to submit changes.
|
|
728
|
+
|
|
729
|
+
## References
|
|
730
|
+
|
|
731
|
+
Brain Memory's architecture and benchmark methodology are grounded in the following work.
|
|
732
|
+
|
|
733
|
+
### Foundations — the agent-memory model Brain implements
|
|
734
|
+
|
|
735
|
+
- [**CoALA — Cognitive Architectures for Language Agents**](https://arxiv.org/abs/2309.02427) (arxiv 2309.02427) — Sumers, Yao, Narasimhan, Griffiths. The agent-memory taxonomy Brain implements directly. Pinned Tier → semantic memory, Skills → procedural memory, session-start aggregator → working memory.
|
|
736
|
+
- [**MemGPT — LLMs as Operating Systems**](https://arxiv.org/abs/2310.08560) (arxiv 2310.08560) — Packer et al. Paging-style memory management.
|
|
737
|
+
- [**Generative Agents — Interactive Simulacra of Human Behavior**](https://arxiv.org/abs/2304.03442) (arxiv 2304.03442) — Park et al. Recency · importance · relevance retrieval blend.
|
|
738
|
+
- [**Memory in the Age of AI Agents**](https://arxiv.org/abs/2512.13564) (arxiv 2512.13564) — Comprehensive survey on agent memory architectures.
|
|
739
|
+
- [**Mem0**](https://arxiv.org/abs/2504.19413) (arxiv 2504.19413) — Human-like memory reinforcement and decay.
|
|
740
|
+
- [**MemOS**](https://arxiv.org/abs/2507.03724) (arxiv 2507.03724) — Memory lifecycle state management.
|
|
741
|
+
|
|
742
|
+
### Memory benchmarks (the suite this work follows)
|
|
743
|
+
|
|
744
|
+
- [**LongMemEval**](https://arxiv.org/abs/2410.10813) (arxiv 2410.10813) — distractor haystacks (S / M / Oracle), abstention category, GPT-4o judge with 97% human agreement.
|
|
745
|
+
- [**MemoryAgentBench**](https://arxiv.org/abs/2507.05257) (arxiv 2507.05257) — four-competency framework; FactConsolidation inspired Scenario C.
|
|
746
|
+
- [**SWE-Bench-CL**](https://arxiv.org/abs/2507.00014) (arxiv 2507.00014) — repo-scoped chronological evaluation; template for Scenario E.
|
|
747
|
+
- [**LoCoMo**](https://arxiv.org/abs/2402.17753) (arxiv 2402.17753) — long-conversation memory benchmark.
|
|
748
|
+
- [**MIRIX**](https://arxiv.org/abs/2507.07957) (arxiv 2507.07957) — realistic synthetic-but-grounded memory benchmarks.
|
|
749
|
+
|
|
750
|
+
### Methodology — judging and benchmark hygiene
|
|
751
|
+
|
|
752
|
+
- [**Preference Leakage in LLM-as-judge**](https://arxiv.org/abs/2502.01534) (arxiv 2502.01534) — drives cross-family judging.
|
|
753
|
+
- [**When Judgment Becomes Noise — position bias**](https://arxiv.org/abs/2509.20293) (arxiv 2509.20293) — drives position-swap mitigation.
|
|
754
|
+
- [**Silent Judge — shortcut bias**](https://arxiv.org/abs/2509.26072) (arxiv 2509.26072) — drives rubric-based judging.
|
|
755
|
+
- [**LastingBench**](https://arxiv.org/abs/2506.21614) (arxiv 2506.21614) — benchmark-leakage defense.
|
|
756
|
+
|
|
757
|
+
### Neuroscience
|
|
758
|
+
|
|
759
|
+
- [**Ebbinghaus forgetting curve**](https://en.wikipedia.org/wiki/Forgetting_curve) — exponential memory decay.
|
|
760
|
+
- [**Spreading activation**](https://en.wikipedia.org/wiki/Spreading_activation) — Collins & Loftus network model of semantic memory.
|
|
761
|
+
- [**Hebbian theory**](https://en.wikipedia.org/wiki/Hebbian_theory) — "neurons that fire together wire together."
|
|
762
|
+
- [**Synaptic homeostasis hypothesis**](https://doi.org/10.1016/j.neuron.2013.10.024) — Tononi & Cirelli's theory of sleep function.
|
|
763
|
+
- [**SM-2 algorithm**](https://en.wikipedia.org/wiki/SuperMemo#Description_of_SM-2_algorithm) — spaced repetition scheduling.
|
|
764
|
+
- [**Memory reconsolidation**](https://en.wikipedia.org/wiki/Memory_consolidation#Reconsolidation) — recalling memories makes them temporarily malleable.
|
|
765
|
+
|
|
766
|
+
### Related projects
|
|
767
|
+
|
|
768
|
+
- [get-shit-done](https://github.com/gsd-build/get-shit-done) — Structured file-based workflow orchestration for AI agents.
|
|
463
769
|
|
|
464
770
|
## License
|
|
465
771
|
|