catlog22/Claude-Code-Workflow
1
0
Fork 0
mirror of https://github.com/catlog22/Claude-Code-Workflow.git synced 2026-02-14 02:42:04 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix: resolve Gemini-identified pipeline data flow inconsistencies

Browse Source 
Fix 3 issues found by Gemini full-chain analysis:

1. (High) cross_cutting_specs path format mismatch: feature-index.json
   uses relative paths ("role/file.md") but context-package uses full
   paths (".workflow/.../role/file.md"). Changed action-planning-agent
   to use endsWith() matching and read from context-package objects
   instead of raw feature-index strings.

2. (Medium) Add explicit feature_index_path field to context-package
   schema in context-search-agent, so task-generate-agent can read it
   directly instead of hardcoding paths. Updated both single-module
   and multi-module prompt templates with fallback logic.

3. (Medium) Expand synthesis_output in context-search-agent to include
   feature_driven sub-object pointing to feature-index.json and
   feature-specs/ directory alongside legacy synthesis-specification.md.
This commit is contained in:
catlog22
2026-02-12 22:06:00 +08:00
parent 99ee4e7d36
commit a169f0e5b8
3 changed files with 35 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
.claude/agents/action-planning-agent.md
View File

@@ -157,9 +157,9 @@ mcp__exa__get_code_context_exa(
- **Load on-demand**: Only load specs referenced by current task's feature mapping
- **Note**: For structured metadata (`feature_id`, `slug`, `priority`), use `feature_index.features[]` instead
- `brainstorm_artifacts.cross_cutting_specs[]`: Cross-cutting concern specifications (from context-package)
- Each spec: `cross_cutting_specs[i]` has `path` and `content`
- Each spec: `cross_cutting_specs[i]` has `path` (full project-relative) and `content`
- **Load on-demand**: Only load when task touches shared/cross-cutting concerns
- **Note**: In feature-index.json, `cross_cutting_specs[]` is a plain string array (relative paths)
- **Path format note**: context-package uses full paths (`.workflow/.../role/file.md`), feature-index.json uses relative paths (`role/file.md`). Match using `endsWith()`.
- `brainstorm_artifacts.role_analyses[]`: Role-specific analyses (legacy fallback, if array not empty)
- Each role: `role_analyses[i].files[j]` has `path` and `content`
- **Only used when**: `feature_index` does not exist (backward compatibility)
@@ -197,11 +197,17 @@ if (contextPackage.brainstorm_artifacts?.feature_index?.exists) {
});
// Step 3: Load cross-cutting specs only when needed
// Note: cross_cutting_specs in feature-index.json is a string array (relative paths)
// Note: feature-index.json uses relative paths ("role/file.md"),
// context-package uses full paths (".workflow/.../role/file.md")
const crossCuttingFromPackage = contextPackage.brainstorm_artifacts.cross_cutting_specs || [];
featureIndex.cross_cutting_specs
.filter(cs => task.context.artifacts.some(a => a.type === 'cross_cutting_spec' && a.path === cs))
.filter(cs => task.context.artifacts.some(a => a.type === 'cross_cutting_spec'))
.forEach(cs => {
const crossCuttingContent = Read(cs);
// Match by path suffix since feature-index uses relative paths
const matched = crossCuttingFromPackage.find(pkg => pkg.path.endsWith(cs));
if (matched) {
const crossCuttingContent = matched.content || Read(matched.path);
}
});
} else if (contextPackage.brainstorm_artifacts?.role_analyses?.length > 0) {