Convert raw plugin analysis results into actionable improvement recommendations for agent-scaffolders and agent-skill-open-specifications. Trigger with "synthesize learnings", "generate improvement recommendations", "what should we improve in our scaffolders", "update our meta-skills based on these findings", or after completing a plugin analysis.
From agent-plugin-analyzernpx claudepluginhub richfrem/agent-plugins-skills --plugin agent-plugin-analyzerThis skill is limited to using the following tools:
acceptance-criteria.mdevals/evals.jsonevals/results.tsvfallback-tree.mdpattern-catalog.mdreferences/acceptance-criteria.mdreferences/fallback-tree.mdreferences/improvement-mapping.mdreferences/input-contract.mdreferences/open-recommendations.mdrequirements.txtCreates consistent pitch decks, one-pagers, investor memos, financial models, accelerator apps, and fundraising materials from a single source of truth.
Provides expertise on electricity/gas procurement, tariff optimization, demand charge management, renewable PPA evaluation, hedging, load profiling, and multi-facility energy strategies.
Provides demand forecasting, safety stock optimization, replenishment planning, and promotional lift estimation for multi-location retailers managing 300-800 SKUs.
This skill requires Python 3.8+ and standard library only. No external packages needed.
To install this skill's dependencies:
pip-compile ./requirements.in
pip install -r ./requirements.txt
See ./requirements.txt for the dependency lockfile (currently empty — standard library only).
Take raw analysis output from analyze-plugin and transform it into concrete, actionable improvements for our meta-skills ecosystem. This is the "close the loop" skill that turns observations into evolution.
Learnings are mapped to three improvement targets:
agent-scaffoldersImprovements to the plugin/skill/hook/sub-agent scaffolding tools.
What to look for:
scaffold.py should supportagent-skill-open-specificationsImprovements to ecosystem standards and authoritative source documentation.
What to look for:
agent-plugin-analyzer (Self-Improvement)Improvements to this analyzer plugin itself.
What to look for:
pattern-catalog.mdoracle-legacy-system-analysis)Improvements to the primary domain plugins in this repository — especially the legacy Oracle Forms/DB analysis plugins.
What to look for:
~~category patterns) for tool-agnostic Oracle analysis workflowsCollect all analysis reports from the current session or from referenced analysis artifacts.
Sort every observation into one of these categories:
| Category | Description | Maps To |
|---|---|---|
| Structural Innovation | Novel directory layouts, component organization | Scaffolders |
| Content Pattern | Reusable content structures (tables, frameworks, checklists) | Specs + Catalog + Domain |
| Execution Pattern | Workflow designs, phase structures, decision trees | Scaffolders + Specs + Domain |
| Integration Pattern | MCP tool usage, connector abstractions, cross-tool design | Specs + Domain |
| Quality Pattern | Testing, validation, compliance approaches | Scaffolders + Specs |
| Meta Pattern | Self-referential or recursive designs (skills that build skills) | Analyzer + Scaffolders |
| Anti-Pattern | Things to avoid, documented pitfalls | Specs |
| Domain Applicability | Patterns transferable to legacy code analysis workflows | Domain |
| Novel Discovery | Something entirely new not in existing catalogs | All targets |
For EACH observation, produce a structured recommendation:
### [Recommendation Title]
**Source**: [Plugin/skill where observed]
**Category**: [from table above]
**Target**: [which meta-skill to improve]
**Priority**: [high / medium / low]
**Observation**: [What was found]
**Current State**: [How our meta-skills handle this today, or "not addressed"]
**Proposed Improvement**: [Specific change to make]
**Example**: [Before/after or concrete illustration]
Rank recommendations by impact:
| Priority | Criteria |
|---|---|
| High | Universal pattern found across many plugins; would improve ALL generated plugins; addresses a gap in current standards |
| Medium | Common pattern found in several plugins; would improve most generated plugins; refines existing standards |
| Low | Niche pattern from specific domain; would improve specialized plugins; nice-to-have enhancement |
Append any newly discovered patterns to references/pattern-catalog.md in the analyze-plugin skill. This is the self-improvement loop — every analysis makes future analyses better.
Append each recommendation to references/open-recommendations.md using this format:
| [YYYY-MM-DD] | [Title] | [Target] | [Priority] | open |
See references/open-recommendations.md for the tracker schema. When a recommendation is
implemented, update its status from open to implemented and add the PR or commit reference.
Format new catalog entries as:
### [Pattern Name]
- **Category**: [Structural / Content / Execution / Integration / Quality / Meta]
- **First Seen In**: [plugin name]
- **Description**: [2-3 sentences]
- **When to Use**: [trigger conditions]
- **Example**: [brief illustration]
Produce a final synthesis report with:
The synthesis report should be a standalone markdown document suitable for:
Iteration Directory Isolation: Do NOT overwrite existing synthesis reports. Always output to a newly isolated directory (e.g. synthesis-reports/run-1/) so historical recommendations are preserved.
Asynchronous Benchmark Metric Capture: Log the total_tokens and duration_ms consumed during the synthesis back to timing.json to track the ROI cost of this meta-analysis.
Always close with a Next Steps section listing the 3 most impactful changes to make first.