Propose improvements or new components for Product Forge
59
42%
Does it follow best practices?
Impact
Pending
No eval scenarios have been run
Passed
No known issues
Optimize this skill with Tessl
npx tessl skill review --optimize ./plugins/claude-code-dev/skills/propose-forge-improvement/SKILL.mdRetrospect on the current session and propose improvements to Product Forge - either enhancing existing components or suggesting new ones.
After working with Product Forge skills, commands, or agents, identify opportunities to improve the ecosystem based on real usage patterns.
/propose-forge-improvement # Analyze and propose improvements
/propose-forge-improvement --save # Save proposal to ~/.claude/learnings/Analyze session for Product Forge opportunities
↓
┌─────────────────────────────────────────────┐
│ Could an existing component be improved? │
│ │
│ Check: │
│ - Skills used → Missing guidance? │
│ - Commands run → Missing options? │
│ - Agents spawned → Missing capabilities? │
│ - Docs referenced → Outdated/incomplete? │
└─────────────────────────────────────────────┘
↓
┌─────────┴─────────┐
↓ ↓
Improvement New Component
to existing proposalWhen the user runs this command:
Scan the session for:
For each identified component, ask:
| Question | If Yes → |
|---|---|
| Missing guidance in a skill? | Skill improvement |
| Missing option in a command? | Command improvement |
| Missing capability in an agent? | Agent improvement |
| Outdated information in docs? | Doc improvement |
| Workflow that could be automated? | New command |
| Knowledge that should be applied automatically? | New skill |
| Complex task needing specialized agent? | New agent |
If proposing an improvement, find the target:
# Search Product Forge for existing components
find ~/.claude/plugins/cache -name "*.md" | xargs grep -l "{component-name}" 2>/dev/nullOr reference known locations:
plugins/{plugin}/skills/{skill-name}/SKILL.mdplugins/{plugin}/commands/{command-name}.mdplugins/{plugin}/agents/{agent-name}.md# Proposed Improvement to Product Forge
## Target Component
**Type**: skill | command | agent | doc
**Location**: plugins/python-experts/skills/django-api/SKILL.md
**Component**: django-api skill
## Current Gap
The django-api skill covers endpoint creation but lacks guidance on:
- Pagination patterns for large datasets
- Cursor-based vs offset pagination trade-offs
- Integration with Django REST Framework pagination classes
## Observed Need
During this session, we implemented pagination for a user listing endpoint.
Had to research DRF pagination classes manually - this knowledge should be
in the skill.
## Suggested Addition
Add a "Pagination Patterns" section:
```python
# Cursor-based pagination (preferred for large datasets)
class UserPagination(CursorPagination):
page_size = 50
ordering = '-created_at'
# Offset pagination (simpler, but slower for deep pages)
class UserPagination(PageNumberPagination):
page_size = 50
max_page_size = 100#### For New Components
```markdown
# Proposed New Component for Product Forge
## Component Type
**Type**: skill | command | agent
**Suggested Name**: migration-safety
**Target Plugin**: python-experts
## Problem Statement
When modifying Django models, there's risk of creating migrations that:
- Lock tables for extended periods
- Cause data loss
- Break backwards compatibility
No current skill covers migration safety patterns.
## Proposed Solution
### If Skill
Create `migration-safety` skill that Claude applies when:
- Modifying Django models
- Creating or reviewing migrations
- Planning database schema changes
Key guidance:
- Additive-only changes for zero-downtime
- Separate deploy for column removal
- Index creation with CONCURRENTLY
### If Command
Create `/check-migration` command that:
- Analyzes pending migrations for safety issues
- Flags risky operations (column drops, type changes)
- Suggests safer alternatives
### If Agent
Create `migration-reviewer` agent that:
- Reviews migration files
- Checks for backwards compatibility
- Suggests deployment order
## Evidence from Session
[What happened in the session that prompted this suggestion]If --save is provided:
Determine type:
improvement → existing component enhancementskill-idea / command-idea / agent-idea → new componentSave to appropriate location:
mkdir -p ~/.claude/learnings/projects/{project-slug}/feedback/{type}/
# Save as: {type}-{timestamp}.mdConfirm:
Proposal saved to ~/.claude/learnings/projects/{project-slug}/feedback/improvement/
Review with: /sync-feedback --reviewNext steps:
[1] Create/modify the component now (I'll help implement)
[2] Save for later review (/propose-forge-improvement --save)
[3] Open GitHub issue (if Product Forge repo accessible)
[4] Dismiss
Select option:Only propose improvements that are:
| Criterion | Description |
|---|---|
| Evidence-based | Rooted in actual session experience |
| Generalizable | Useful beyond this specific project |
| Non-trivial | Meaningful improvement, not minor tweaks |
| Actionable | Clear enough to implement |
| Scoped | One focused improvement per proposal |
When multiple opportunities exist, prioritize:
/propose-project-learning which targets project CLAUDE.md/sync-feedback for batch review and export0ebe7ae
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.