opencode-review-helper 0.3.0 → 0.3.1

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 +72 -40
  2. package/package.json +1 -1
package/README.md CHANGED
@@ -21,10 +21,9 @@ Restart opencode after setup to load the plugin. The plugin uses `@latest` so it
21
21
 
22
22
  ## Usage
23
23
 
24
- ### Slash Command: `/review-order`
25
-
26
- The quickest way to use this plugin:
24
+ **Direct Slash Command Use**
27
25
 
26
+ Run the slash command:
28
27
  ```
29
28
  /review-order
30
29
  ```
@@ -36,14 +35,74 @@ This runs the full workflow:
36
35
 
37
36
  Output is a unified table showing all files to review, with changed files first (prioritized by dependencies and impact) followed by external files that may need attention.
38
37
 
39
- ### Agent: `review-helper`
38
+ **Agent Use**
39
+
40
+ Tab twice to go to Review-Helper mode:
41
+ <img width="1968" height="188" alt="image" src="https://github.com/user-attachments/assets/1a548c1a-b81d-4982-a2db-c2a1525a6d7a" />
42
+
43
+ Enter your prompt:
44
+ ```
45
+ Take a look at everything since the 0.2.0 release as if it were a single PR. Guide me through how to review it.
46
+ ```
40
47
 
41
- The plugin provides a `review-helper` agent that orchestrates two subagents:
48
+ Output:
49
+ ```
50
+ Review Order for v0.2.0 → HEAD
51
+ | # | File | Reason | Score |
52
+ |---|------|--------|-------|
53
+ | 1 | src/tools/review-order.ts | Core scoring/sorting logic; foundation for all agents | 80 |
54
+ | 2 | src/agents/review-order.ts | Sub-agent wrapping review_order tool | 70 |
55
+ | 3 | src/agents/impact-explorer.ts | Sub-agent wrapping impact_analysis tool | 70 |
56
+ | 4 | src/agents/review-helper.ts | Orchestrator coordinating sub-agents | 60 |
57
+ | 5 | src/agents/index.ts | Barrel export for agent definitions | 50 |
58
+ | 6 | src/index.ts | Plugin entry point; command registration | 40 |
59
+ | 7 | package.json | Version bump 0.2.0 → 0.3.0 | 30 |
60
+ | 8 | AGENTS.md | Local testing/release docs | 20 |
61
+ | 9 | README.md | User-facing docs reorganization | 10 |
62
+ | 10 | agents/*.md (deleted) | Removed markdown agent files | 5 |
63
+ ---
64
+ Summary of Changes
65
+ Theme: Subagent architecture for /review-order slash command
66
+ Structural Changes
67
+ - Added: src/agents/ directory with 4 new files
68
+ - Deleted: agents/*.md files (agent prompts moved to TypeScript)
69
+ - Refactored: src/index.ts significantly simplified (136 lines → ~30 lines of logic)
70
+ Key Review Points
71
+ | File | What to Check |
72
+ |------|---------------|
73
+ | src/tools/review-order.ts | +139 lines. New scoring algorithm? Dependency detection? |
74
+ | src/agents/review-helper.ts | Orchestration flow: review-order → impact → merge |
75
+ | src/index.ts | Verify plugin registration still works after major trim |
76
+ | agents/*.md deletion | Confirm prompts weren't lost, just moved to .ts |
77
+ Testing Gaps
78
+ ⚠️ No test files for any changed code. Manual testing via /review-order command required.
79
+ ---
80
+ Want me to read specific files for deeper analysis?
81
+ ```
42
82
 
43
- - `review-helper:review-order` - Determines optimal file review order
44
- - `review-helper:impact-explorer` - Finds external code impacted by changes
83
+ ## Agents
45
84
 
46
- The agent does NOT make up its own recommendations or analysis—it only presents what the tools return. It also does NOT automatically run tests or apply fixes.
85
+ Three agents available for direct use or extension:
86
+
87
+ | Agent | Purpose |
88
+ |-------|---------|
89
+ | `review-helper` | Orchestrator - runs full `/review-order` workflow |
90
+ | `review-helper:review-order` | Determines optimal file review order |
91
+ | `review-helper:impact-explorer` | Finds external impacted code |
92
+
93
+ The orchestrator does NOT make up its own recommendations—it only presents what the subagents return. It also does NOT automatically run tests or apply fixes.
94
+
95
+ ### Examples
96
+
97
+ Run impact analysis only:
98
+ ```bash
99
+ opencode run --agent review-helper:impact-explorer "Analyze impact of current changes"
100
+ ```
101
+
102
+ Get review order without impact merge:
103
+ ```bash
104
+ opencode run --agent review-helper:review-order "Get review order for changed files"
105
+ ```
47
106
 
48
107
  ## Configuration
49
108
 
@@ -57,12 +116,10 @@ Create `~/.config/opencode/review-helper.json` or `.opencode/review-helper.json`
57
116
  "review_order": {
58
117
  "type_priority": [
59
118
  "migration",
60
- "schema",
119
+ "schema",
61
120
  "model",
62
121
  "service",
63
- "resolver",
64
- "component",
65
- "test"
122
+ "..."
66
123
  ],
67
124
  "instructions": "Always review database changes before business logic"
68
125
  },
@@ -74,44 +131,19 @@ Create `~/.config/opencode/review-helper.json` or `.opencode/review-helper.json`
74
131
  }
75
132
  ```
76
133
 
134
+ *Default `type_priority` includes 18 keywords. See [src/config.ts](src/config.ts) for full list.*
135
+
77
136
  ### Configuration Options
78
137
 
79
138
  | Option | Description | Default |
80
139
  |--------|-------------|---------|
81
140
  | `models.explorer` | Model for sub-agent exploration | `google/gemini-3-flash` |
82
- | `review_order.type_priority` | File type keywords in priority order | See above |
141
+ | `review_order.type_priority` | File type keywords in priority order | See config.ts |
83
142
  | `review_order.instructions` | Custom ordering instructions | - |
84
143
  | `impact_analysis.max_depth` | Transitive analysis depth (1-3) | `2` |
85
144
  | `impact_analysis.max_results_per_level` | Max results per category | `50` |
86
145
  | `impact_analysis.exclude_patterns` | Glob patterns to skip | Test files |
87
146
 
88
- ## Tools
89
-
90
- The plugin also exposes two tools directly:
91
-
92
- ### `review_order`
93
-
94
- Suggests optimal order to review changed files.
95
-
96
- Arguments:
97
- - `files` (optional): Specific files to analyze. If omitted, uses git diff.
98
- - `instructions` (optional): Custom ordering instructions
99
- - `impact_data` (optional): Impact analysis results to merge external files into the list
100
-
101
- Output:
102
- - Prioritized file list with rationale
103
- - Dependency graph showing import relationships
104
- - Scores based on: type priority, dependents count, complexity, external impact
105
-
106
- ### `impact_analysis`
107
-
108
- Finds code outside the changeset that could break.
109
-
110
- Output:
111
- - Direct consumers (files that import changed code)
112
- - Transitive impact (files that use files that use changed code)
113
- - Test coverage gaps (changed files without tests)
114
-
115
147
  ## License
116
148
 
117
149
  MIT
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "opencode-review-helper",
3
- "version": "0.3.0",
3
+ "version": "0.3.1",
4
4
  "description": "OpenCode plugin for reviewing AI-generated code changes - suggests review order and analyzes impact",
5
5
  "type": "module",
6
6
  "main": "dist/index.js",