src/autoskillit/skills_extended/elaborate-phase/SKILL.md
Elaborate a migration plan phase into a complete self-contained implementation plan. Use when user says "elaborate phase", "elaborate phase N", or "phase elaboration". Assesses codebase, writes detailed phase plan, then validates with dry walkthrough.
npx skillsauth add talont-org/autoskillit elaborate-phaseInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Elaborate a single phase from a high-level migration plan into a complete, self-contained implementation plan. The plan should contain all background context needed for implementation.
NEVER:
Phase#.md files unless you have a specific question that requires looking up a detailrun_in_background: true is prohibited)ALWAYS:
{plan_dir}/Phase#.md)/autoskillit:dry-walkthrough on the written Phase#.md to validateAll plans live in a dedicated directory:
{plan_directory}/
├── {master_plan_name}.md # Unique name (e.g., feature_migration_plan.md)
├── Phase0.md # Elaborated phase 0 (if exists)
├── Phase1.md # Elaborated phase 1 (if exists)
├── Phase2.md # etc.
└── ...
Phase#.md, contain detailed implementation instructionsWhen elaborating the first phase:
.md files exist in the directoryUser message example:
Here's the plan directory: {plan_dir}/ with master plan {master_plan}.md. Elaborate phase 0.
When elaborating subsequent phases:
Phase#.md files - assess from codebase directlyUser message example:
Same directory. Phase 0 is implemented. Elaborate phase 1.
Extract from user message:
Phase#.md files)Load and understand:
IMPORTANT: Do NOT read previous Phase#.md files. They exist in the directory but reading them wastes context. Only look up a specific detail from a previous phase if you encounter a concrete question during assessment that cannot be answered from the codebase itself.
Launch parallel Explore subagents to understand the current state:
Subagent 1: Affected Files Assessment
- Read all files that will be touched by this phase
- Document their current state (structure, dependencies, imports)
- Identify any deviations from what the master plan expected
Subagent 2: Dependency Analysis
- Trace imports and dependencies for affected files
- Identify all consumers of code being moved/changed
- Map the dependency graph
Subagent 3: Test Coverage
- Find tests related to affected code
- Understand what test changes will be needed
- Identify any integration tests that may break
Subagent 4: Pattern Discovery
- Search for similar patterns in the codebase
- Identify conventions that should be followed
- Note any existing utilities that should be reused
Subagent 5: State Verification (for subsequent phases)
- Verify previous phases were implemented correctly
- Check that expected files/directories exist
- Confirm imports were updated as planned
Output assessment summary to terminal:
## Codebase Assessment Complete
### Current State
- {component}: {current status}
- {file}: {current location and state}
### Key Findings
- {finding that affects the plan}
- {deviation from master plan expectations}
### Dependencies Mapped
- {affected file} imports: {list of imports}
- {affected file} imported by: {list of consumers}
Create a complete implementation plan at the output location. The plan should include all necessary background context.
Plan Structure:
# Phase {N}: {Phase Name}
## Executive Summary
{2-3 sentences describing what this phase accomplishes and why}
{Any critical changes from the original master plan, with rationale}
---
## Prerequisites
Before starting Phase {N}:
- [ ] {Prerequisite check}
- [ ] {Required state verification}
- [ ] Previous phases implemented
- [ ] All tests pass: run the project's test suite
- [ ] No uncommitted changes: `git status`
---
## Background Context
{Everything the implementer needs to know that isn't in other sections}
### Relevant Code Locations
| Component | Current Location | Purpose |
|-----------|------------------|---------|
| {component} | {path} | {brief description} |
### Key Dependencies
{Description of important dependencies and why they matter}
---
## Phase {N} Tasks
### Task {N}.1: {Task Name}
{Description of what this task accomplishes}
**Files to Modify:**
- `{path}`: {what changes}
**Implementation Steps:**
1. {Specific action with details}
2. {Specific action with details}
**Verification:**
```bash
{command to verify task completion}
{Same structure as above}
After completing Phase {N}, verify:
| Path | Type | Purpose | |------|------|---------| | {path} | {Created/Modified/Moved/Deleted} | {purpose} |
Phase {N} explicitly avoids:
{Project-specific guidelines relevant to this phase}
Phase {N+1}: {Name}
{Brief description of what comes next, so implementer understands boundaries}
### Step 5: Dry Walkthrough
Invoke `/autoskillit:dry-walkthrough` on the `Phase{N}.md` file just written.
### Step 6: Update Master Plan (If Required)
If the dry walkthrough reveals issues that affect subsequent phases:
1. Update the master plan to reflect corrections
2. Report changes to terminal
### Step 7: Report to Terminal
After writing the plan, output summary:
Phase: {N} - {Phase Name} Plan Directory: {plan_directory} Output: Phase{N}.md Status: {READY / REVISED from master plan}
{List any modifications made during elaboration}
{If master plan was updated due to downstream impact}
Ready for implementation.
## Output Location
Phase plans are written to the same directory as the master plan:
{plan_directory}/Phase{N}.md
Where `{N}` is replaced with the phase number (0, 1, 2, etc.).
Example directory after elaborating phases 0 and 1:
/path/to/plans/ ├── feature_migration_plan.md # Master plan ├── Phase0.md # Elaborated phase 0 └── Phase1.md # Elaborated phase 1
## Backward Compatibility Policy
**Backward compatibility code is only acceptable when:**
1. There is an explicit cleanup phase in the migration plan
2. The cleanup phase explicitly lists removal of the compatibility code
3. The compatibility code is clearly marked for removal
**If not planned for cleanup, do NOT add:**
- Import shims or re-exports for old paths
- Deprecation warnings
- Compatibility adapters
- Fallback mechanisms
## Example Usage
### First Phase Example
**User:** "Here's the plan directory: /path/to/plans/ with master plan feature_migration_plan.md. Elaborate phase 0."
**Process:**
1. Read master plan from `/path/to/plans/feature_migration_plan.md`
2. No existing `Phase#.md` files (first phase)
3. Launch subagents to assess current codebase state
4. Write elaborated plan to `/path/to/plans/Phase0.md`
5. Run `/autoskillit:dry-walkthrough` on `Phase0.md`
6. Update master plan if dry walkthrough reveals downstream issues
7. Report findings and status
### Subsequent Phase Example
**User:** "Same directory. Phase 0 is implemented. Elaborate phase 1."
**Process:**
1. Read master plan from `/path/to/plans/feature_migration_plan.md`
2. Launch subagents to assess:
- Current state (post Phase 0 implementation)
- Files touched by Phase 1
- Dependencies and consumers
3. Write elaborated plan to `/path/to/plans/Phase1.md`
4. Run `/autoskillit:dry-walkthrough` on `Phase1.md`
5. Update master plan if dry walkthrough reveals downstream issues
6. Report findings and status
## Related Skills
- **`/autoskillit:dry-walkthrough`** - Used internally for validation
- **`/autoskillit:make-plan`** - Creates new plans (this skill elaborates existing ones)
- **`/autoskillit:implement-worktree`** - Implements elaborated plans
development
Generate YAML recipes for .autoskillit/recipes/. Use when user says "make script skill", "generate script", "script a workflow", "write a script", "create a script", "new recipe", "write a pipeline", or when loaded by other skills for script formatting.
data-ai
Create Uncertainty Representation visualization planning spec showing error bar definitions, distribution-aware alternatives, and multi-seed variance protocols. Statistical lens answering "How is uncertainty honestly represented?"
data-ai
Create Temporal Dynamics visualization planning spec showing axis scaling (linear vs log), smoothing disclosure, epoch/step alignment, run aggregation (mean + variance bands), early-stopping markers, and wall-clock vs step-count x-axis. Temporal lens answering "Are training dynamics shown clearly and honestly?"
data-ai
Create Narrative Story Arc visualization planning spec showing visual consistency across the report (same color = same model everywhere), logical figure progression, redundant figure detection, and narrative dependency between figures. Narrative lens answering "Do the figures tell a coherent story across the report?"