vibe-forge 0.8.1 → 0.8.2

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.
Files changed (51) hide show
  1. package/.claude/commands/configure-vcs.md +102 -102
  2. package/.claude/commands/forge.md +218 -218
  3. package/.claude/hooks/worker-loop.js +220 -217
  4. package/.claude/settings.json +89 -89
  5. package/README.md +149 -191
  6. package/agents/aegis/personality.md +303 -303
  7. package/agents/anvil/personality.md +278 -278
  8. package/agents/architect/personality.md +260 -260
  9. package/agents/crucible/personality.md +362 -362
  10. package/agents/crucible-x/personality.md +210 -210
  11. package/agents/ember/personality.md +293 -293
  12. package/agents/flux/personality.md +248 -248
  13. package/agents/furnace/personality.md +342 -342
  14. package/agents/herald/personality.md +249 -249
  15. package/agents/oracle/personality.md +284 -284
  16. package/agents/pixel/personality.md +140 -140
  17. package/agents/planning-hub/personality.md +473 -473
  18. package/agents/scribe/personality.md +253 -253
  19. package/agents/slag/personality.md +268 -268
  20. package/agents/temper/personality.md +270 -270
  21. package/bin/cli.js +372 -372
  22. package/bin/forge-daemon.sh +477 -477
  23. package/bin/forge-setup.sh +662 -661
  24. package/bin/forge-spawn.sh +164 -164
  25. package/bin/forge.sh +566 -566
  26. package/docs/commands.md +8 -8
  27. package/package.json +77 -77
  28. package/{bin → src}/lib/agents.sh +177 -177
  29. package/{bin → src}/lib/check-aliases.js +50 -50
  30. package/{bin → src}/lib/colors.sh +45 -44
  31. package/{bin → src}/lib/config.sh +347 -347
  32. package/{bin → src}/lib/constants.sh +241 -241
  33. package/{bin → src}/lib/daemon/budgets.sh +107 -107
  34. package/{bin → src}/lib/daemon/dependencies.sh +146 -146
  35. package/{bin → src}/lib/daemon/display.sh +128 -128
  36. package/{bin → src}/lib/daemon/notifications.sh +273 -273
  37. package/{bin → src}/lib/daemon/routing.sh +93 -93
  38. package/{bin → src}/lib/daemon/state.sh +163 -163
  39. package/{bin → src}/lib/daemon/sync.sh +103 -103
  40. package/{bin → src}/lib/database.sh +357 -357
  41. package/{bin → src}/lib/frontmatter.js +106 -106
  42. package/{bin → src}/lib/heimdall-setup.js +113 -113
  43. package/{bin → src}/lib/heimdall.js +265 -265
  44. package/src/lib/index.sh +25 -0
  45. package/{bin → src}/lib/json.sh +264 -264
  46. package/{bin → src}/lib/terminal.js +452 -452
  47. package/{bin → src}/lib/util.sh +126 -126
  48. package/{bin → src}/lib/vcs.js +349 -349
  49. package/{context → templates}/project-context-template.md +122 -122
  50. package/config/task-template.md +0 -159
  51. package/config/templates/handoff-template.md +0 -40
package/README.md CHANGED
@@ -1,252 +1,210 @@
1
- # Vibe Forge ⚒️
1
+ # Vibe Forge
2
2
 
3
3
  A multi-agent development orchestration system for terminal-native vibe coding.
4
4
 
5
- ## Vision
5
+ ## What Is This?
6
6
 
7
- Vibe Forge transforms your terminal into a collaborative AI development environment. Multiple Claude agents - each with distinct personalities and specializations - work together to build software, coordinated through a file-based task system.
7
+ Vibe Forge transforms your terminal into a collaborative AI development environment. Multiple Claude agents, each with distinct personalities and specializations, work together to build software. You talk to a Planning Hub that coordinates the team, then spawn workers into separate terminal tabs to execute in parallel.
8
8
 
9
9
  ```
10
- ┌─────────────────────────────────────────────────────────────────┐
11
- │ PLANNING HUB │
12
- │ (Your main terminal session) │
13
- │ │
14
- │ You + Architect · Aegis · Ember · Pixel · Oracle
15
- └─────────────────────────────────────────────────────────────────┘
16
-
17
-
18
- ┌─────────────────────────────────────────────────────────────────┐
19
- │ PLANNING HUB 🔥 │
20
- │ Task Distribution & Orchestration │
21
- └─────────────────────────────────────────────────────────────────┘
22
-
23
- ┌────────────┬───────┴───────┬────────────┐
24
- ▼ ▼ ▼ ▼
25
- ┌─────────┐ ┌─────────┐ ┌─────────┐ ┌─────────┐
26
- │ Anvil │ │ Furnace │ │Crucible │ │Sentinel │
27
- │ 🔨 │ │ 🔥 │ │ 🧪 │ │ 🛡️ │
28
- │Frontend │ │ Backend │ │ Testing │ │ Review │
29
- └─────────┘ └─────────┘ └─────────┘ └─────────┘
10
+ YOU
11
+ |
12
+ /forge plan
13
+ |
14
+ +-----------+-----------+
15
+ | PLANNING HUB |
16
+ | Oracle Architect |
17
+ | Aegis Pixel |
18
+ | Ember Crucible |
19
+ +-----------+-----------+
20
+ |
21
+ /forge spawn <agent>
22
+ |
23
+ +--------+-------+-------+--------+
24
+ | | | | |
25
+ Anvil Furnace Crucible Temper Scribe
26
+ FE BE QA Review Docs
30
27
  ```
31
28
 
32
- ## Key Features
29
+ ## Install
33
30
 
34
- - **Personality-driven agents** - Each agent has a distinct voice, expertise, and decision-making style
35
- - **File-based task coordination** - Reliable, debuggable, no WebSocket complexity
36
- - **Token-efficient design** - Context stored locally, minimal wire traffic
37
- - **Terminal-native** - Built for Windows Terminal, works with any terminal supporting tabs
31
+ ```bash
32
+ npx vibe-forge init
33
+ ```
38
34
 
39
- ## Agents
35
+ This sets up Vibe Forge in your project: detects your platform, configures your terminal, creates project context files, and installs the `/forge` slash command into Claude Code.
40
36
 
41
- ### Core Agents (Always Running)
37
+ **Prerequisites:** [Claude Code CLI](https://claude.ai/download), Node.js 18+, Git
42
38
 
43
- | Agent | Icon | Role |
44
- |-------|------|------|
45
- | Planning Hub | 🔥 | Chief Orchestrator - distributes tasks, tracks progress |
46
- | Sentinel | 🛡️ | Code Reviewer - quality gates, adversarial review |
39
+ ## Quick Start
47
40
 
48
- ### Worker Agents (Per-Task)
41
+ ```bash
42
+ # Start the Planning Hub (multi-expert planning session)
43
+ /forge
49
44
 
50
- | Agent | Icon | Role |
51
- |-------|------|------|
52
- | Anvil | 🔨 | Frontend Dev - components, UI, styling |
53
- | Furnace | 🔥 | Backend Dev - APIs, database, services |
54
- | Crucible | 🧪 | Tester/QA - tests, bug hunting |
55
- | Scribe | 📜 | Documentation - docs, README, API specs |
56
- | Herald | 📯 | Release Manager - versioning, deployment |
45
+ # Or jump straight to planning a feature
46
+ /forge plan user authentication
57
47
 
58
- ### Planning Hub Voices (Your Terminal)
48
+ # Spawn a worker agent in a new terminal tab
49
+ /forge spawn anvil
59
50
 
60
- The Planning Hub is a **multi-voice planning session** -- not separate agents. When you run `/forge`, these expert voices collaborate in a single session:
51
+ # Check what's happening
52
+ /forge status
61
53
 
62
- | Voice | Icon | Expertise |
63
- |-------|------|-----------|
64
- | Forge Master | 🔥 | Orchestration & task coordination |
65
- | Architect | 🏛️ | Technical design & architecture |
66
- | Aegis | 🔒 | Security & risk |
67
- | Ember | ⚙️ | DevOps & infrastructure |
68
- | Pixel | 🎨 | User experience & design |
69
- | Oracle | 📊 | Requirements & scope |
70
- | Crucible | 🧪 | Quality & edge cases |
54
+ # See all commands
55
+ /forge help
56
+ ```
71
57
 
72
- ### Specialists (On-Demand)
58
+ ## Agents
73
59
 
74
- | Agent | Icon | Role |
75
- |-------|------|------|
76
- | Architect | 🏛️ | System Architect |
77
- | Ember | ⚙️ | DevOps/Infrastructure |
78
- | Aegis | 🔒 | Security Specialist |
79
- | Pixel | 🎨 | UX Designer |
80
- | Oracle | 🔮 | Product Owner / Requirements |
81
- | Loki | 🎭 | Lateral Thinker, Assumption Challenger |
60
+ ### Planning Hub (Your Terminal)
82
61
 
83
- ### Red Team (On-Demand)
62
+ When you run `/forge`, a multi-voice planning session starts. These expert voices collaborate to help you scope, design, and decompose work:
84
63
 
85
- | Agent | Icon | Role |
86
- |-------|------|------|
87
- | Slag | 💀 | Red Team Lead - OWASP, auth attacks, engagement reporting |
88
- | Flux | | Red Team Operator - dependency CVEs, CI/CD, supply chain |
64
+ | Voice | Icon | Speaks When |
65
+ |-------|------|-------------|
66
+ | Forge Master | :fire: | Tasks, assignments, workflow, coordination |
67
+ | Architect | :classical_building: | Architecture, patterns, tech decisions |
68
+ | Aegis | :shield: | Auth, security, vulnerabilities |
69
+ | Ember | :gear: | DevOps, CI/CD, deployment |
70
+ | Pixel | :art: | UX, user flows, accessibility |
71
+ | Oracle | :bar_chart: | Requirements, scope, priorities |
72
+ | Crucible | :test_tube: | Edge cases, test strategy, quality |
73
+ | Loki | :performing_arts: | Challenges assumptions, lateral thinking |
89
74
 
90
- ## Project Structure
75
+ ### Worker Agents (Separate Terminals)
91
76
 
92
- ```
93
- vibe-forge/
94
- ├── agents/ # Agent definitions
95
- │ ├── planning-hub/
96
- │ │ ├── personality.md # Identity, voice, principles
97
- │ │ ├── capabilities.md # Commands, tools, decisions
98
- │ │ └── context-template.md # Session startup context
99
- │ ├── sentinel/
100
- │ ├── anvil/
101
- │ ├── furnace/
102
- │ ├── crucible/
103
- │ └── ...
104
- ├── tasks/ # Task lifecycle folders
105
- │ ├── pending/ # New tasks waiting for pickup
106
- │ ├── in-progress/ # Currently being worked on
107
- │ ├── completed/ # Done, ready for review
108
- │ ├── review/ # Under Sentinel review
109
- │ ├── approved/ # Passed review
110
- │ ├── needs-changes/ # Review feedback to address
111
- │ └── merged/ # Archive
112
- ├── specs/ # Planning documents
113
- │ ├── epics/
114
- │ └── stories/
115
- ├── context/ # Shared context files
116
- │ ├── project-context.md # Tech stack, patterns, rules
117
- │ └── forge-state.yaml # Current forge status
118
- └── config/ # Configuration
119
- ├── agents.json # Agent roster (source of truth)
120
- ├── agent-manifest.yaml # Agent documentation (non-normative)
121
- ├── task-template.md # Task file template
122
- └── task-types.yaml # Task routing rules
123
- ```
77
+ Spawn these into new terminal tabs to execute tasks:
124
78
 
125
- ## Task Lifecycle
79
+ | Agent | Aliases | Role |
80
+ |-------|---------|------|
81
+ | anvil | frontend, ui, fe | Frontend Developer |
82
+ | furnace | backend, api, be | Backend Developer |
83
+ | crucible | test, qa | Tester / QA |
84
+ | scribe | docs, documentation | Documentation |
85
+ | herald | release, deploy | Release Manager |
86
+ | ember | devops, infra | DevOps Engineer |
126
87
 
127
- ```
128
- ┌─────────┐ ┌─────────────┐ ┌───────────┐ ┌────────┐
129
- │ pending │ -> │ in-progress │ -> │ completed │ -> │ review │
130
- └─────────┘ └─────────────┘ └───────────┘ └────────┘
131
-
132
- ┌──────────────┐ │
133
- │ needs-changes│ <────────────────┤
134
- └──────────────┘ │
135
- │ │
136
- ▼ ▼
137
- ┌─────────────┐ ┌──────────┐
138
- │ in-progress │ │ approved │
139
- └─────────────┘ └──────────┘
140
-
141
-
142
- ┌─────────┐
143
- │ merged │
144
- └─────────┘
145
- ```
88
+ ### Review Agents
146
89
 
147
- ## Getting Started
90
+ | Agent | Aliases | Role |
91
+ |-------|---------|------|
92
+ | temper | review, cr | Code Reviewer (compliance + correctness) |
93
+ | crucible-x | adversarial, cx | Adversarial Reviewer (tries to break it) |
148
94
 
149
- ### Prerequisites
95
+ ### Specialists
150
96
 
151
- - Claude Code CLI ([install](https://claude.ai/download))
152
- - Windows Terminal (recommended) or any terminal with tabs
153
- - Node.js 16+ (for npx installer)
154
- - Git
97
+ | Agent | Aliases | Role |
98
+ |-------|---------|------|
99
+ | architect | arch, sage | System Architect |
100
+ | aegis | security, sec | Security Specialist |
101
+ | pixel | ux, ui-design | UX Designer |
102
+ | oracle | product, po | Product Owner |
103
+ | loki | brainstorm, contrarian | Assumption Challenger |
155
104
 
156
- ### Quick Start
105
+ ### Red Team
157
106
 
158
- ```bash
159
- # In your project directory
160
- npx vibe-forge init
161
- ```
107
+ | Agent | Aliases | Role |
108
+ |-------|---------|------|
109
+ | slag | redteam, pentest | Red Team Lead |
110
+ | flux | infra-sec, chaos | Infrastructure Security |
162
111
 
163
- This will:
112
+ ## How It Works
164
113
 
165
- 1. Clone Vibe Forge into `_vibe-forge/`
166
- 2. Detect your platform and terminal
167
- 3. Set up the daemon and configuration
168
- 4. Create a project context file
114
+ ### Planning Mode
169
115
 
170
- Then start the Planning Hub:
116
+ When you describe a goal, the Hub enters a 4-phase planning flow:
171
117
 
172
- ```bash
173
- cd _vibe-forge
174
- ./bin/forge.sh
175
- ```
118
+ 1. **Discovery** - Oracle asks clarifying questions about users, goals, constraints
119
+ 2. **Decomposition** - Architect breaks the goal into epics with success metrics
120
+ 3. **Tasking** - Forge Master creates stories and tasks, assigns to agents
121
+ 4. **Commit** - Epic and task files written to disk, ready for workers
176
122
 
177
- ### Manual Setup
123
+ ### Task System
178
124
 
179
- If you prefer not to use npx:
125
+ Tasks flow through folders on disk. No database required.
180
126
 
181
- ```bash
182
- # Clone into your project
183
- git clone https://github.com/sugar-crash-studios/vibe-forge.git _vibe-forge
127
+ ```
128
+ pending/ -> in-progress/ -> completed/ -> review/ -> approved/ -> merged/
129
+ |
130
+ needs-changes/ (back to worker)
131
+ ```
184
132
 
185
- # Run setup
186
- cd _vibe-forge
187
- ./bin/forge-setup.sh
133
+ Workers pick up tasks from `pending/` on startup. Temper reviews completed work. The daemon routes tasks automatically.
188
134
 
189
- # Start the Planning Hub
190
- ./bin/forge.sh
191
- ```
135
+ ### Daemon
192
136
 
193
- ### Updating
137
+ An optional background daemon monitors the forge:
194
138
 
195
139
  ```bash
196
- npx vibe-forge update
140
+ ./bin/forge.sh daemon start # Start background monitoring
141
+ ./bin/forge.sh daemon status # Check what's happening
142
+ ./bin/forge.sh daemon stop # Stop the daemon
197
143
  ```
198
144
 
199
- ## Slash Commands
145
+ The daemon provides:
146
+ - Task routing between folders
147
+ - Agent status tracking
148
+ - Token budget warnings for long-running agents
149
+ - Dependency resolution (respects `blocked_by` in task files)
150
+ - Attention notifications when agents need help
200
151
 
201
- When using Claude Code inside your project, use the `/forge` command:
152
+ ### Dashboard
202
153
 
203
- ```
204
- /forge - Start the Planning Hub (default)
205
- /forge status - Show status dashboard
206
- /forge spawn <agent> - Spawn worker in new terminal
207
- /forge task [desc] - Create a new task
208
- /forge redteam [scope] - Launch red team engagement
209
- /forge help - Show available commands
210
- ```
154
+ A web dashboard at `http://localhost:2800` shows:
155
+ - Task counts and status
156
+ - Agent activity feed
157
+ - Issue detection
158
+ - Real-time updates via WebSocket
211
159
 
212
- Agents (with aliases):
160
+ ### Per-Project Customization
213
161
 
214
- | Agent | Aliases | Role |
215
- |-------|---------|------|
216
- | anvil | frontend, ui, fe | Frontend Developer |
217
- | furnace | backend, api, be | Backend Developer |
218
- | crucible | test, testing, qa | Tester / QA |
219
- | sentinel | review, reviewer, cr | Code Reviewer |
220
- | scribe | docs, documentation | Documentation |
221
- | herald | release, deploy | Release Manager |
222
- | ember | devops, ops, infra | DevOps |
223
- | aegis | security, sec, appsec | Security |
224
- | slag | redteam, pentest | Red Team Lead |
225
- | flux | infra-sec, chaos | Red Team Operator |
162
+ Add agent-specific rules for your project in `context/agent-overrides/`:
226
163
 
227
- Use either the forge name or any alias: `/forge spawn frontend` or `/forge spawn anvil`
164
+ ```markdown
165
+ <!-- context/agent-overrides/anvil.md -->
166
+ - Use Tailwind CSS, no custom CSS files
167
+ - All components in src/components/ with PascalCase
168
+ - Use shadcn/ui for base components
169
+ ```
228
170
 
229
- Red team engagements: `/forge redteam [scope]`
171
+ These rules are injected into the agent's system prompt at spawn time.
230
172
 
231
- ## Token Efficiency
173
+ ## Project Structure
232
174
 
233
- Vibe Forge is designed for minimal token usage:
175
+ ```
176
+ your-project/
177
+ _vibe-forge/
178
+ agents/ # Agent personalities (16 agents)
179
+ bin/ # CLI, daemon, dashboard, spawn scripts
180
+ config/ # agents.json, task templates
181
+ context/ # Project context, agent overrides
182
+ specs/ # Epics and stories
183
+ tasks/ # Task lifecycle folders
184
+ docs/ # Security, architecture, agent docs
185
+ ```
234
186
 
235
- 1. **Local context** - Agents read from files, not conversation history
236
- 2. **Task files as truth** - Instructions in files, not repeated in chat
237
- 3. **Reference, don't duplicate** - Point to paths, don't paste contents
238
- 4. **Batch updates** - One status report per cycle, not per task
239
- 5. **Exception-based** - Report problems, not smooth operations
187
+ ## Commands Reference
240
188
 
241
- ## Philosophy
189
+ | Command | Description |
190
+ |---------|-------------|
191
+ | `/forge` | Start the Planning Hub |
192
+ | `/forge plan <feature>` | Plan a feature with the full team |
193
+ | `/forge status` | Show status dashboard |
194
+ | `/forge spawn <agent>` | Spawn worker in new terminal |
195
+ | `/forge task [desc]` | Create a new task |
196
+ | `/forge redteam [scope]` | Launch red team engagement |
197
+ | `/forge help` | Show all commands |
242
198
 
243
- > "A forge is not a factory. Each piece is crafted with intention."
199
+ ## Security
244
200
 
245
- Vibe Forge embraces the craft of software development. Each agent brings expertise and personality to their work. The goal isn't maximum automation - it's maximum collaboration between human and AI.
201
+ Vibe Forge uses a defense-in-depth permission model:
246
202
 
247
- ## Acknowledgments
203
+ 1. **Allowlist** (`.claude/settings.json`) - Pre-approves safe operations
204
+ 2. **Heimdall** (pre-tool hook) - Enforces forge policies (branch protection, naming)
205
+ 3. **Claude Code prompts** - Anything not allowlisted still requires approval
248
206
 
249
- Inspired by BMAD (Business Model-Agnostic Development) methodology and its multi-agent workflow system.
207
+ See [docs/security.md](docs/security.md) for the full security model.
250
208
 
251
209
  ## License
252
210