@proxysoul/soulforge 1.0.3 → 1.1.0

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 (3) hide show
  1. package/README.md +27 -22
  2. package/dist/index.js +917 -774
  3. package/package.json +1 -1
package/README.md CHANGED
@@ -2,14 +2,14 @@
2
2
 
3
3
  <p align="center">
4
4
  <strong>Graph-Powered Code Intelligence</strong><br/>
5
- Multi-agent coding with codebase-aware AI — embedded Neovim, 22 themes, terminal-native
5
+ Multi-agent coding with codebase-aware AI — embedded Neovim, 21 themes, terminal-native
6
6
  </p>
7
7
 
8
8
  <p align="center">
9
9
  <a href="LICENSE"><img src="https://img.shields.io/badge/License-BSL%201.1-blue.svg" alt="License" /></a>
10
10
  <a href="#"><img src="https://img.shields.io/badge/version-1.0.0-brightgreen.svg" alt="Version" /></a>
11
11
  <a href="https://www.typescriptlang.org/"><img src="https://img.shields.io/badge/TypeScript-strict-blue.svg" alt="TypeScript" /></a>
12
- <a href="#testing"><img src="https://img.shields.io/badge/tests-2292%20passing-brightgreen.svg" alt="Tests" /></a>
12
+ <a href="#testing"><img src="https://img.shields.io/badge/tests-2296%20passing-brightgreen.svg" alt="Tests" /></a>
13
13
  <a href="https://bun.sh"><img src="https://img.shields.io/badge/runtime-Bun-f472b6.svg" alt="Bun" /></a>
14
14
  </p>
15
15
 
@@ -169,7 +169,7 @@ Installable skill system for domain-specific capabilities. Destructive action ap
169
169
 
170
170
  ## Architecture
171
171
 
172
- The Forge Agent is the orchestrator. It holds 34 tools including the `dispatch` tool, which creates an AgentBus and launches parallel subagents. Subagents share file/tool caches through the bus and coordinate edits via ownership tracking.
172
+ The Forge Agent is the orchestrator. It holds 33 tools including the `dispatch` tool, which creates an AgentBus and launches parallel subagents. Subagents share file/tool caches through the bus and coordinate edits via ownership tracking.
173
173
 
174
174
  ```mermaid
175
175
  graph TB
@@ -177,7 +177,7 @@ graph TB
177
177
  Chat --> Forge[Forge Agent]
178
178
 
179
179
  subgraph Forge Tools
180
- Tools[34 Direct Tools]
180
+ Tools[33 Tools]
181
181
  Dispatch[dispatch tool]
182
182
  end
183
183
 
@@ -186,11 +186,11 @@ graph TB
186
186
 
187
187
  Dispatch --> |creates| Bus[AgentBus<br/>file cache · tool cache<br/>findings · edit ownership]
188
188
 
189
- Bus --> |spawns with bus| E1[Explore Agent]
190
- Bus --> |spawns with bus| E2[Explore Agent]
191
- Bus --> |spawns with bus| C[Code Agent]
189
+ Bus --> |spawns| I[Investigate Agent<br/>broad analysis]
190
+ Bus --> |spawns| E[Explore Agent<br/>targeted extraction]
191
+ Bus --> |spawns| C[Code Agent<br/>edits]
192
192
 
193
- E1 & E2 & C --> |read/write cache| Bus
193
+ I & E & C --> |read/write cache| Bus
194
194
 
195
195
  Tools --> Intel[Intelligence Router]
196
196
  Tools --> Nvim[Neovim<br/>msgpack-RPC]
@@ -240,20 +240,20 @@ sequenceDiagram
240
240
  participant F as Forge Agent
241
241
  participant D as dispatch tool
242
242
  participant B as AgentBus
243
- participant E1 as Explore 1
244
- participant E2 as Explore 2
243
+ participant I as Investigate
244
+ participant E as Explore
245
245
  participant C as Code Agent
246
246
 
247
247
  U->>F: "Audit auth and refactor middleware"
248
248
  F->>D: dispatch(3 tasks)
249
249
  D->>B: new AgentBus()
250
250
  par Concurrent (3 slots, staggered)
251
- D->>E1: launch (t=0ms)
252
- D->>E2: launch (t=150ms)
253
- D->>C: launch (dependsOn: E1, E2)
251
+ D->>I: launch investigate (t=0ms)
252
+ D->>E: launch explore (t=150ms)
253
+ D->>C: launch code (dependsOn: I, E)
254
254
  end
255
- E1->>B: cache file reads + findings
256
- E2->>B: cache file reads + findings
255
+ I->>B: soul_grep counts + soul_analyze + findings
256
+ E->>B: targeted reads + findings
257
257
  Note over B: Code agent reads<br/>hit cache instantly
258
258
  B-->>C: dependencies resolved
259
259
  C->>C: Edit with full context
@@ -411,7 +411,7 @@ soulforge --version # Version info
411
411
 
412
412
  ### Slash Commands
413
413
 
414
- 85 commands available — press `/` or `Ctrl+K` to browse. Key ones by category:
414
+ 86 commands available — press `/` or `Ctrl+K` to browse. Key ones by category:
415
415
 
416
416
  **Models & Providers**
417
417
  `/model` `/router` `/provider` `/model-scope`
@@ -457,17 +457,18 @@ soulforge --version # Version info
457
457
 
458
458
  ## Tool Suite
459
459
 
460
- SoulForge ships 34 tools organized by capability:
460
+ SoulForge ships 33 tools organized by capability:
461
461
 
462
462
  ### Code Intelligence
463
463
 
464
464
  | Tool | What it does |
465
465
  |------|-------------|
466
- | `read_code` | Extract function/class/type by name (LSP-powered) |
466
+ | `read_file` | Read files with optional `target` param for symbol extraction (LSP-powered) |
467
467
  | `navigate` | Definition, references, call hierarchy, implementations |
468
468
  | `analyze` | File diagnostics, unused symbols, complexity |
469
469
  | `rename_symbol` | Compiler-guaranteed rename across all files |
470
470
  | `move_symbol` | Move to another file + update all importers |
471
+ | `rename_file` | Rename/move a file with import path updates |
471
472
  | `refactor` | Extract function/variable, organize imports |
472
473
 
473
474
  ### Codebase Analysis (zero LLM cost)
@@ -503,6 +504,10 @@ SoulForge ships 34 tools organized by capability:
503
504
 
504
505
  **Planning:** `plan`, `update_plan_step`, `task_list`, `ask_user`
505
506
 
507
+ **Discovery:** `discover_pattern`, `test_scaffold`, `skills`
508
+
509
+ **Tool Management:** `request_tools`, `release_tools` (agent-managed mode — disabled by default)
510
+
506
511
  </details>
507
512
 
508
513
  ---
@@ -723,7 +728,7 @@ See [GETTING_STARTED.md](GETTING_STARTED.md) for the full reference.
723
728
  ## Testing
724
729
 
725
730
  ```bash
726
- bun test # 2292 tests across 49 files
731
+ bun test # 2296 tests across 49 files
727
732
  bun run typecheck # tsc --noEmit
728
733
  bun run lint # biome check (lint + format)
729
734
  bun run lint:fix # auto-fix
@@ -735,7 +740,7 @@ bun run lint:fix # auto-fix
735
740
 
736
741
  | Document | Description |
737
742
  |----------|-------------|
738
- | **[Command Reference](docs/commands-reference.md)** | All 85 commands by category |
743
+ | **[Command Reference](docs/commands-reference.md)** | All 86 commands by category |
739
744
  | **[Headless Mode](docs/headless.md)** | Non-interactive CLI for CI/CD, scripting, automation |
740
745
  | **[Architecture](docs/architecture.md)** | System overview, data flow, component lifecycle |
741
746
  | **[Repo Map](docs/repo-map.md)** | PageRank, cochange, blast radius, clone detection |
@@ -766,7 +771,7 @@ sf --headless CLI mode — CI/CD, scripts, automation ✓ shipped
766
771
  SoulForge TUI Full experience (what you're looking at now)
767
772
  ```
768
773
 
769
- - **`@soulforge/intelligence`** — graph intelligence, 34 tools, and agent orchestration as an importable package. Build your own AI tools on top of SoulForge's brain.
774
+ - **`@soulforge/intelligence`** — graph intelligence, 33 tools, and agent orchestration as an importable package. Build your own AI tools on top of SoulForge's brain.
770
775
  - **`@soulforge/mcp`** — expose soul_grep, soul_find, soul_analyze, soul_impact, navigate, read_code as MCP tools. Any AI tool that supports MCP gets SoulForge's graph intelligence.
771
776
  - **`sf --headless`** — non-interactive mode. Pipe in a prompt, get back results. For CI/CD, automation, and benchmarks. [Documentation →](docs/headless.md)
772
777
 
@@ -775,7 +780,7 @@ SoulForge TUI Full experience (what you're looking at now)
775
780
  - **Repo Map visualization** — interactive dependency graph, PageRank heatmap, blast radius explorer
776
781
  - **GitHub CLI integration** — native `gh_pr`, `gh_issue`, `gh_status` tools with structured output
777
782
  - **Dispatch worktrees** — git worktree per code agent for conflict-free parallel edits
778
- - **[ACP support](https://agentclientprotocol.com/)** — Agent Client Protocol integration. Run SoulForge as a coding agent inside Zed, JetBrains, Neovim (agentic.nvim), or any ACP-compatible editor. Headless mode already covers 80% of the protocol surface — `sf --acp` would expose graph intelligence, multi-agent dispatch, and all 34 tools via JSON-RPC 2.0 over stdio
783
+ - **[ACP support](https://agentclientprotocol.com/)** — Agent Client Protocol integration. Run SoulForge as a coding agent inside Zed, JetBrains, Neovim (agentic.nvim), or any ACP-compatible editor. Headless mode already covers 80% of the protocol surface — `sf --acp` would expose graph intelligence, multi-agent dispatch, and all 33 tools via JSON-RPC 2.0 over stdio
779
784
 
780
785
  **Planned:**
781
786
  - **Monorepo graph support** — cross-package dependency tracking for pnpm/npm/yarn workspaces, Cargo workspaces, Go workspaces (`go.work`), Nx/Turborepo, and Bazel/Buck. Currently the repo map treats each workspace root as an isolated unit — cross-package imports resolve as external dependencies instead of internal edges. This means PageRank, blast radius, and unused export detection don't span package boundaries.