zam-core 0.9.1 β†’ 0.9.2

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
@@ -1,183 +1,141 @@
1
1
  # ZAM (Zusammen) 🀝
2
2
 
3
- **The Symbiotic Learning Kernel: Elevating Human Intelligence through AI Collaboration.**
3
+ **Do real work with your AI β€” and keep the knowledge instead of losing it.**
4
4
 
5
5
  > *ZAM is Bavarian for "together".*
6
6
 
7
- ZAM is an open-source framework that creates a deep symbiosis between humans and AI. While conventional AI often renders people passive, ZAM harnesses technological progress to deepen human knowledge, prevent cognitive decline caused by automation, and organically connect communities.
7
+ ZAM turns everyday work with your AI agent into active-recall practice, so growing
8
+ automation doesn't mean growing dependence. You get the task done **and** you get
9
+ sharper β€” the two stop being a trade-off.
8
10
 
9
- ---
10
-
11
- ## πŸš€ Phase 1: Individual Symbiosis *(Current Focus)*
12
-
13
- Phase 1 centers on the interaction between a person and their personal agent. The goal: solve tasks efficiently while actively acquiring and retaining valuable knowledge.
14
-
15
- > **"Don't just automate β€” Elevate."**
16
-
17
- When the AI takes on tasks, the ZAM Learning Kernel ensures the human stays in control and in the loop:
11
+ Don't just automate. **Elevate.**
18
12
 
19
- - **Knowledge Retention** β€” Through Active Recall and Spaced Repetition, ZAM reminds you of the domain knowledge you need for your work, so that growing automation doesn't mean growing dependence.
20
- - **Competence Transfer** β€” While ZAM handles the routine (emails, contact lists, scheduling), your agent simultaneously teaches you concepts from modern project management, volunteer coordination, or whatever domain you're working in.
21
- - **Time Reinvestment** β€” The time you gain is not consumed by passivity, but reinvested in learning new, more challenging skills.
22
-
23
- ### Example: Organizing a Parish Festival
24
-
25
- | Step | What ZAM does | What you do |
26
- |---|---|---|
27
- | **Planning** | Structures the agenda and timeline | Define the vision and priorities |
28
- | **Learning** | Teaches you effective PR and volunteer motivation | Engage with the concepts |
29
- | **Automation** | Drafts invitations, manages documents | Focus on human connection |
13
+ 🌐 **[zam-os.org](https://zam-os.org)** β€” the project website, in 7 languages.
30
14
 
31
15
  ---
32
16
 
33
- ## 🌐 Phase 2: The Connected Community *(Outlook)*
17
+ ## Who ZAM is for
34
18
 
35
- Once your personal agent knows your knowledge level, interests, and growth goals, ZAM expands to the community layer.
19
+ Anyone who works with an AI agent and doesn't want to get rusty as it takes on more β€”
20
+ whether you're learning the field (say, a *Fachinformatiker* apprentice building durable
21
+ competence), sharpening your craft on the job, or simply keen to keep growing while you
22
+ automate. If you're pairing with Claude, Codex, Copilot & co. anyway, ZAM makes that time
23
+ compound into lasting skill.
36
24
 
37
- ### Agent-to-Agent Marketplace
25
+ ---
38
26
 
39
- Agents communicate with each other to match community needs and offerings:
27
+ ## What ZAM does today
40
28
 
41
- 1. **Need Broadcasting** β€” Your agent signals: *"We're looking for a cook and a musician for the parish festival."*
42
- 2. **Growth-Aware Matching** β€” Other agents scan their people's profiles β€” not just for availability, but for learning potential:
43
- > *"My person wants to practice cooking at scale (learning goal) β€” the festival is the perfect opportunity for active knowledge application."*
44
- 3. **Accredited Communities** β€” Matching is prioritized within trusted circles (e.g. your own parish) to foster genuine human encounters.
29
+ - **Rides along in your AI agent.** As you work a real task, ZAM breaks it into small
30
+ knowledge concepts, notices which ones you're due to revisit, and weaves them into
31
+ the session.
32
+ - **Watches you work.** Do a step well on your own and ZAM quietly marks it learned β€” no
33
+ interruption. When no real task can show what you know, it asks a focused recall
34
+ question. Both are active recall.
35
+ - **Remembers what you're forgetting.** Every concept is scheduled with FSRS-5 spaced
36
+ repetition over a prerequisite graph, so ZAM resurfaces things right before they'd slip.
37
+ - **Stays on your machine.** One local SQLite database (`~/.zam/zam.db`), shared by the
38
+ agent and the Desktop Studio. Review works offline; local LLMs (Ollama, FastFlowLM)
39
+ are supported.
45
40
 
46
41
  ---
47
42
 
48
- ## πŸ›  Technical Architecture: The Learning Kernel
43
+ ## Two places to use ZAM
49
44
 
50
- ZAM is designed as an **AI-agnostic kernel** β€” a CLI tool that integrates seamlessly into existing workflows:
45
+ Your **agent app** is the main workbench. **ZAM Desktop Studio** is for setup, content,
46
+ and focused review. They share the same local database, so progress in one shows up in
47
+ the other.
51
48
 
52
- - **CLI Integration** β€” Compatible with `Claude Code`, `Codex`, `Copilot CLI`, and `Gemini CLI`.
53
- - **Modularity** β€” The system can be forked for region- or culture-specific adaptations (*Social Forking*).
49
+ ### 1. In your AI agent β€” *the workbench*
54
50
 
55
- ### Two Repositories, One System
51
+ This is where the real learning happens: turning actual tasks into practice, observing
52
+ your work, and guiding you step by step. ZAM connects to the agent apps you already use:
56
53
 
57
- ZAM is split into two concerns:
54
+ | Agent | Connect with |
55
+ |---|---|
56
+ | **Claude** (Code / desktop app) | `zam agent connect claude-code` |
57
+ | **Codex** | `zam agent connect codex` |
58
+ | **Antigravity** | `zam agent connect antigravity` |
59
+ | **OpenCode** | `zam agent connect opencode` |
60
+ | **GitHub Copilot** | `zam agent connect copilot` |
61
+ | **Goose** | `zam agent connect goose` |
58
62
 
59
- - **Core** ([`zam-os/zam`](https://github.com/zam-os/zam)) β€” The AI-agnostic learning kernel, CLI, bridge protocol, and system beliefs. Shared by everyone.
60
- - **Personal** (fork of [`zam-os/zam-personal`](templates/personal/)) β€” Your beliefs, your goals, your identity. You fork it, you own it.
61
- Get started: `zam whoami --set <your-id>`
63
+ One command writes the MCP config (your agent may ask you to approve the server). Then
64
+ just type **`/zam`** β€” or say "let's do this together with ZAM" β€” and work normally.
62
65
 
63
- ---
66
+ ### 2. ZAM Desktop Studio β€” *setup, content & graph*
64
67
 
65
- ## πŸ€– Agent & MCP Integration
68
+ A native app (`zam ui`) for the things a chat window isn't good at:
66
69
 
67
- ZAM integrates directly with your coding agents (such as Claude Code, Codex, Antigravity, or OpenCode) via the **Model Context Protocol (MCP)**. This lets your agent drive ZAM sessions automatically using tool calls.
70
+ - **Easier configuration** β€” pick your language and local AI model in a settings panel,
71
+ not a config file.
72
+ - **Import your own material** β€” paste notes, point ZAM at a source, or walk a structured
73
+ curriculum; a guided wizard turns any of them into review cards.
74
+ - **Edit your content** β€” a real editor for concepts, questions, and prerequisites.
75
+ - **See your knowledge graph** β€” your concepts as a living map of what builds on what.
76
+ - **Review** β€” run focused active-recall rounds right in the app.
68
77
 
69
- To connect your agent harness:
70
78
  ```bash
71
- # Connect to your agent client (claude-code, antigravity, codex, or opencode)
72
- zam agent connect claude-code
79
+ zam ui # launch the Studio
80
+ zam ui --build # one-time: build a native installer (needs Rust)
73
81
  ```
74
- This writes the necessary MCP server config. Your harness may still ask you to approve the server or its tools. You can also print the config changes:
75
- ```bash
76
- zam agent connect claude-code --print
77
- ```
78
-
79
- Once connected, your agent will use MCP tools (`zam_status`, `zam_get_reviews`, `zam_session_start`, etc.) with the bridge CLI as a fallback.
80
-
81
- ---
82
-
83
- ## 🧹 Review maintenance
84
-
85
- Review sessions are not limited to `1`-`4` recall ratings anymore. When a card is wrong, obsolete, or unwanted, the review flow can now:
86
-
87
- - edit token fields inline
88
- - deprecate the token
89
- - hard-delete the token after an impact preview + confirmation
90
- - delete only your personal card while keeping the token
91
-
92
- The same maintenance actions are also available from the CLI:
93
-
94
- - `zam token edit --slug <slug> ...`
95
- - `zam token delete --slug <slug>` for preview, then `--force` to delete
96
- - `zam card delete --user <id> --token <slug>`
97
- - `zam bridge review-action ...` for AI clients
98
82
 
99
- Token deletion is global. Card deletion is per-user.
83
+ > **Review works in both places.** Observation and guided task-work happen inside your agent.
100
84
 
101
85
  ---
102
86
 
103
- ## πŸ”„ Keeping ZAM up to date
87
+ ## Quickstart
104
88
 
105
- Check whether a newer release is out, then update β€” ZAM picks the right mechanism for how this copy was installed:
89
+ **1. Get ZAM.** Grab an installer from [Releases](https://github.com/zam-os/zam/releases),
90
+ or build from source:
106
91
 
107
92
  ```bash
108
- zam update # apply the latest release (asks first; -y to skip)
109
- zam update check # only report whether an update is available
93
+ git clone https://github.com/zam-os/zam.git && cd zam
94
+ npm install && npm run build
110
95
  ```
111
96
 
112
- What `zam update` does per install channel:
97
+ **2. Set up β€” one guided wizard:**
113
98
 
114
- - **Developer** (source checkout, the default for contributors) β€” pulls the latest source, reinstalls dependencies, rebuilds the CLI, and refreshes the skill files (`zam setup --force`) in the current instance. Restart your agent client (e.g. Claude Code) afterwards to load the refreshed `/zam` skill.
115
- - **winget / Homebrew** β€” defers to `winget upgrade` / `brew upgrade`, so a package-managed install is never self-replaced.
116
- - **Direct download / desktop** β€” applies a signed in-place update through ZAM Desktop.
99
+ ```bash
100
+ zam init
101
+ ```
117
102
 
118
- `zam update` refuses to touch a developer checkout with uncommitted changes; commit or stash them first, or pass `--force`. See the [β€œApproachable Setup and Self-Update” ADR](docs/adr/2026-06-13b-approachable-setup-and-self-update.md) for the design.
103
+ `zam init` picks a workspace, detects your hardware, offers to install a local AI runner
104
+ (Ollama or FastFlowLM), initializes your database, and wires the `/zam` skill.
119
105
 
120
- ---
106
+ **3. Connect your agent:**
121
107
 
122
- ## πŸ› Vision: A Flourishing Future
108
+ ```bash
109
+ zam agent connect claude-code # or codex Β· antigravity Β· opencode Β· copilot Β· goose
110
+ ```
123
111
 
124
- ZAM is a tool for the transition to a world where care and shared growth are the common currency.
112
+ **4. Learn while you work.** Open your agent, start a real task, and type **`/zam`**. It
113
+ checks what's due, plans the concepts behind the task, hands you the work, watches how it
114
+ goes, and updates your schedule.
125
115
 
126
- - **Resource Stewardship** β€” Agents help manage community finances (e.g. a 10% solidarity model) and optimize collective purchasing.
127
- - **Human Proximity** β€” Technology steps back to enable genuine person-to-person exchange.
128
- - **Global Scalability** β€” Supported by institutions like the global church, ZAM aims to become a standard for a just, educated, and caring world community.
116
+ Prefer a gentler start? Run `zam ui`, import your training material, and do a review round
117
+ in the Studio.
129
118
 
130
119
  ---
131
120
 
132
- ## πŸ–₯ ZAM Desktop Studio
133
-
134
- ZAM now includes a cross-platform desktop application inside the [`desktop/`](desktop/) directory. Built with **Tauri v2**, **Vite**, **TypeScript**, and **Vanilla CSS**, it provides a premium dark-mode dashboard and active-recall learning studio that securely shares the same SQLite database as the CLI.
135
-
136
- ### How to Run on Windows, macOS, or Linux:
137
-
138
- 1. **Build the CLI Bridge**:
139
- Ensure you have compiled the latest CLI code in the repository root:
140
- ```bash
141
- npm install
142
- npm run build
143
- ```
144
-
145
- 2. **Launch the Desktop Application**:
146
- Navigate to the `desktop/` directory, install dependencies, and run Tauri:
147
- ```bash
148
- cd desktop
149
- npm install
150
- npm run tauri dev
151
- ```
121
+ ## How it works
152
122
 
153
- This will compile the secure Rust backend, spin up the Vite development server, and open the native desktop window. For detailed settings, key bindings, and local AI (Ollama/FastFlowLM) configurations, see the [Desktop README](desktop/README.md).
123
+ - **Token** β€” one atomic concept worth remembering, tagged with a Bloom level (1 remember β†’ 5 create).
124
+ - **Card** β€” your personal spaced-repetition state for a token (FSRS-5).
125
+ - **Prerequisites** β€” a graph of what must be understood first; ZAM won't quiz a concept
126
+ whose foundations you've just forgotten.
127
+ - **Sessions** β€” every work/learning episode is logged, so ratings come from real evidence.
154
128
 
155
- ### πŸ“¦ Automated GitHub Releases & Native Installers
156
-
157
- We have integrated a **GitHub Actions CI/CD workflow** that currently compiles and packages native installers for Windows (`.msi`/`.exe`) and Linux (`.deb`/`.rpm`) when a new git tag is pushed. AppImage packaging is temporarily deferred because `linuxdeploy` cannot process the bundled native `libsql` module reliably. macOS builds remain supported from source, but signed/notarized macOS release artifacts are deferred until the Apple signing account is available.
158
-
159
- To release a new version (e.g., `v0.1.0`):
160
-
161
- ```bash
162
- # Tag the current commit
163
- git tag v0.1.0
164
-
165
- # Push the tag to GitHub
166
- git push origin v0.1.0
167
- ```
168
-
169
- This automatically spins up the Windows and Linux compiler environments on GitHub and creates a drafted GitHub Release containing their native redistributables.
129
+ The learning engine is an **AI-agnostic kernel** with zero LLM dependencies; the agent
130
+ layer just drives it. See [Architecture](docs/ARCHITECTURE.md).
170
131
 
171
132
  ---
172
133
 
173
- ## πŸ“– Documentation
134
+ ## Documentation
174
135
 
175
- - [Deutsche Version](README.de.md)
176
- - [Contributing Guide](CONTRIBUTING.md)
177
- - [Architecture](docs/ARCHITECTURE.md)
178
-
179
- ---
136
+ - 🌐 [zam-os.org](https://zam-os.org) β€” project website, in 7 languages
137
+ - [Usage & maintenance](docs/USAGE.md) Β· [Contributing](CONTRIBUTING.md) Β· [Architecture](docs/ARCHITECTURE.md)
180
138
 
181
- ## πŸ“„ License
139
+ ## License
182
140
 
183
- Apache 2.0 License β€” see [LICENSE](LICENSE) for details.
141
+ Apache 2.0 β€” see [LICENSE](LICENSE).
package/dist/cli/app.js CHANGED
@@ -9038,7 +9038,7 @@ async function readWebLink(url) {
9038
9038
  redirect: "manual",
9039
9039
  signal: controller.signal,
9040
9040
  headers: {
9041
- "User-Agent": "ZAM-Content-Studio/0.9.1"
9041
+ "User-Agent": "ZAM-Content-Studio/0.9.2"
9042
9042
  }
9043
9043
  });
9044
9044
  if (res.status >= 300 && res.status < 400) {