@audaxic/memory-claude 1.0.4 → 1.0.5

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.
Files changed (2) hide show
  1. package/README.md +15 -13
  2. package/package.json +1 -1
package/README.md CHANGED
@@ -11,6 +11,7 @@ npx @audaxic/memory-claude install
11
11
  This will:
12
12
  1. Prompt for your Audaxic API key (saved to `~/.config/audaxic/config.json`)
13
13
  2. Copy the plugin to `~/.claude/skills/audaxic-memory/`
14
+ 3. Configure the Audaxic MCP server (`.mcp.json` with your API key)
14
15
 
15
16
  The plugin auto-loads on your next Claude Code session.
16
17
 
@@ -18,30 +19,31 @@ Get your API key from: https://memory-dashboard.audaxic.com
18
19
 
19
20
  ## How It Works
20
21
 
21
- The plugin registers two Claude Code hooks:
22
+ The plugin combines three mechanisms:
22
23
 
23
- ### `UserPromptSubmit` → `inject.js`
24
+ ### Skill → auto-search on questions
24
25
 
25
- When you submit a prompt, this hook:
26
- 1. Buffers your prompt to the plugin data directory
27
- 2. Searches Audaxic Memory for relevant context (`POST /search`)
28
- 3. Injects retrieved memories as `<audaxic_memory_context>` into Claude via `additionalContext`
26
+ An Agent Skill (`skills/audaxic-memory/SKILL.md`) instructs Claude to search Audaxic Memory before answering questions about your past decisions, projects, and preferences — then answer directly from the results.
29
27
 
30
- ### `Stop` → `capture.js`
28
+ ### MCP server → trusted memory tools
31
29
 
32
- When Claude finishes responding, this hook:
33
- 1. Buffers the assistant's response
34
- 2. When the buffer reaches 30 messages, flushes to `POST /add`
35
- 3. Skips captures from automatic continuations (`stop_hook_active`)
30
+ The plugin bundles the Audaxic MCP server (`.mcp.json`), giving Claude `search_memories` / `add_memory` / `get_memories` tools. MCP tool results are trusted by Claude, so memory recall actually reaches the answer.
31
+
32
+ ### Hooks → background saving
33
+
34
+ - `UserPromptSubmit` → `inject.mjs`: buffers your prompt and injects context via `additionalContext` (a passive enhancement — injection may be deprioritized by Claude Code, which is why the skill + MCP path exists)
35
+ - `Stop` → `capture.mjs`: buffers the assistant's response; flushes to `POST /add` at 30 messages
36
36
 
37
37
  ## API Key
38
38
 
39
- The plugin reads your API key in this order:
39
+ The plugin uses your API key in this order:
40
40
 
41
41
  1. `CLAUDE_PLUGIN_OPTION_API_TOKEN` — set via the plugin's `userConfig` prompt during `/plugin install`
42
42
  2. `~/.config/audaxic/config.json` — shared config used by all Audaxic integrations
43
43
 
44
- Both hooks fail silently if no key is found — they never block Claude Code.
44
+ The key is also written into the plugin's `.mcp.json` at install time (mode 0600) so the MCP server authenticates automatically.
45
+
46
+ All scripts fail silently if no key is found — they never block Claude Code.
45
47
 
46
48
  ## Architecture
47
49
 
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@audaxic/memory-claude",
3
- "version": "1.0.4",
3
+ "version": "1.0.5",
4
4
  "description": "Claude Code plugin for Audaxic Memory — automatically saves conversation facts and injects semantic context",
5
5
  "type": "module",
6
6
  "bin": {