@cwim/kanban 1.1.15 → 1.1.17

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 +60 -8
  2. package/package.json +1 -1
package/README.md CHANGED
@@ -2,7 +2,9 @@
2
2
 
3
3
  > Your AI's long-term memory. Visualized.
4
4
 
5
- **CWIM Kanban gives Claude Code a persistent memory layer.** It remembers what you were working on, recalls context automatically, and shows you everything on a live dashboard.
5
+ **CWIM Kanban gives your AI agent a persistent memory layer.** It remembers what you were working on, recalls context automatically, and shows you everything on a live dashboard.
6
+
7
+ Works with **Claude Code** and **OpenCode**.
6
8
 
7
9
  No more "what were we doing again?" between sessions.
8
10
 
@@ -38,8 +40,13 @@ kanban
38
40
  ```bash
39
41
  # Start the dashboard
40
42
  kanban
43
+ ```
44
+
45
+ ### Claude Code Setup
41
46
 
42
- # Add to your Claude Code MCP config (~/.claude/claude.json)
47
+ Add to your Claude Code MCP config (`~/.claude/claude.json`):
48
+
49
+ ```json
43
50
  {
44
51
  "mcpServers": {
45
52
  "kanban": {
@@ -50,9 +57,26 @@ kanban
50
57
  }
51
58
  ```
52
59
 
53
- ## Making Claude Use It
60
+ ### OpenCode Setup
61
+
62
+ Add to your OpenCode config (`opencode.json` in project root or `~/.config/opencode/opencode.json`):
63
+
64
+ ```json
65
+ {
66
+ "mcp": {
67
+ "kanban": {
68
+ "command": "npx",
69
+ "args": ["@cwim/kanban", "mcp"]
70
+ }
71
+ }
72
+ }
73
+ ```
74
+
75
+ ## Making Your AI Agent Use It
54
76
 
55
- Just installing the MCP isn't enough — Claude needs instructions to use it.
77
+ Just installing the MCP isn't enough — your AI agent needs instructions to use it.
78
+
79
+ ### For Claude Code
56
80
 
57
81
  Add a `CLAUDE.md` file to your project root:
58
82
 
@@ -70,6 +94,7 @@ Use the cwim-kanban MCP to track all work in this project.
70
94
  ### Rules
71
95
  - Always check for existing tasks before creating new ones
72
96
  - **One task per unit of work** - If a request involves multiple distinct steps (e.g., "fix auth and update docs"), create separate tasks for each step instead of one combined task
97
+ - **Always verify the active session before creating tasks** - Call `session_list` first, confirm the active session matches the current project, and call `session_switch` if it doesn't
73
98
  - Use tags consistently (e.g., "bug", "feature", "refactor", "docs")
74
99
  - Append notes liberally - they build context for future sessions
75
100
  - Move tasks to "blocked" immediately when stuck, with explanation
@@ -78,6 +103,33 @@ Use the cwim-kanban MCP to track all work in this project.
78
103
 
79
104
  This makes the behavior automatic — no need to ask Claude every session.
80
105
 
106
+ ### For OpenCode
107
+
108
+ Add an `AGENTS.md` file to your project root (or add to your existing one):
109
+
110
+ ```markdown
111
+ ## Task Tracking
112
+ Use the cwim-kanban MCP to track all work in this project.
113
+
114
+ ### Workflow
115
+ 1. **Before starting**: Call `task_recall` with what you're about to work on
116
+ 2. **Starting a task**: Create or move to `in-progress`
117
+ 3. **Making progress**: Append notes with discoveries, decisions, or blockers
118
+ 4. **Finishing**: Move to `done` and append a summary note
119
+ 5. **Blocked**: Move to `blocked` with a note explaining why
120
+
121
+ ### Rules
122
+ - Always check for existing tasks before creating new ones
123
+ - **One task per unit of work** - If a request involves multiple distinct steps (e.g., "fix auth and update docs"), create separate tasks for each step instead of one combined task
124
+ - **Always verify the active session before creating tasks** - Call `session_list` first, confirm the active session matches the current project, and call `session_switch` if it doesn't
125
+ - Use tags consistently (e.g., "bug", "feature", "refactor", "docs")
126
+ - Append notes liberally - they build context for future sessions
127
+ - Move tasks to "blocked" immediately when stuck, with explanation
128
+ - Keep task titles concise but descriptive
129
+ ```
130
+
131
+ This makes the behavior automatic for OpenCode sessions as well.
132
+
81
133
  ## Memory Features
82
134
 
83
135
  ### Smart Context Recall
@@ -105,11 +157,11 @@ Each note is timestamped and preserved. The task grows smarter as you work.
105
157
 
106
158
  ### Session Isolation
107
159
 
108
- Each Claude Code project gets its own memory space. Work on multiple projects without context bleeding:
160
+ Each project gets its own memory space. Work on multiple projects without context bleeding:
109
161
 
110
- - Auto-detected from `~/.claude/projects/`
162
+ - Auto-detected from `~/.claude/projects/` (Claude Code) and `~/.config/opencode/` (OpenCode)
111
163
  - Switch between sessions via dashboard, CLI, or MCP
112
- - "Independent Mode" for non-Claude work
164
+ - "Independent Mode" for non-AI work
113
165
 
114
166
  ### Keyword Search
115
167
 
@@ -209,7 +261,7 @@ Claude Code → MCP Server (stdio) → session tasks.json ← HTTP Server ← Da
209
261
  ## Requirements
210
262
 
211
263
  - Node.js 18+
212
- - Claude Code (optional - dashboard works independently)
264
+ - Claude Code or OpenCode (optional - dashboard works independently)
213
265
 
214
266
  ## License
215
267
 
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@cwim/kanban",
3
- "version": "1.1.15",
3
+ "version": "1.1.17",
4
4
  "description": "Minimal Kanban task tracking with MCP integration for Claude Code",
5
5
  "type": "module",
6
6
  "main": "dist/index.js",