Gap Reconnaissance Skill
LIBRARY-FIRST PROTOCOL (MANDATORY)
Before writing ANY code, you MUST check:
Step 1: Library Catalog
- Location:
.claude/library/catalog.json
- If match >70%: REUSE or ADAPT
Step 2: Patterns Guide
- Location:
.claude/docs/inventories/LIBRARY-PATTERNS-GUIDE.md
- If pattern exists: FOLLOW documented approach
Step 3: Existing Projects
- Location:
D:\Projects\*
- If found: EXTRACT and adapt
Decision Matrix
| Match |
Action |
| Library >90% |
REUSE directly |
| Library 70-90% |
ADAPT minimally |
| Pattern exists |
FOLLOW pattern |
| In project |
EXTRACT |
| No match |
BUILD (add to library after) |
Metadata
- Name: gap-reconnaissance
- Version: 1.0.0
- Category: research
- Triggers: ["gap analysis", "find solution", "reconnaissance", "research github", "what framework", "what library"]
- Priority: HIGH
Purpose
Systematic process for identifying gaps in a project, researching solutions, extracting reusable patterns, and creating integration guides - then cleaning up to minimize storage.
The 7-Phase Process
+-----------------------------------------------------------------------+
| GAP RECONNAISSANCE WORKFLOW |
+-----------------------------------------------------------------------+
| |
| PHASE 1: GAP IDENTIFICATION |
| +------------------------------------------------------------------+ |
| | - Analyze project architecture docs | |
| | - Identify missing components (100% gaps) | |
| | - Determine criticality (P0/P1/P2) | |
| | - Document gap with clear requirements | |
| +------------------------------------------------------------------+ |
| | |
| v |
| PHASE 2: TECHNICAL CRITERIA |
| +------------------------------------------------------------------+ |
| | - Define minimum viable solution | |
| | - Check tech stack compatibility | |
| | - Language match (Python/Rust/TypeScript) | |
| | - Deployment match (Docker/K8s/serverless) | |
| | - Integration points (APIs, data formats) | |
| | - Estimate integration effort | |
| | - Define success metrics | |
| +------------------------------------------------------------------+ |
| | |
| v |
| PHASE 3: GITHUB RESEARCH |
| +------------------------------------------------------------------+ |
| | - WebSearch for "[gap] framework 2025 github code" | |
| | - WebSearch for "[gap] paper 2025 arxiv implementation" | |
| | - Filter by: | |
| | - Stars (>1k preferred) | |
| | - Last commit (<6 months) | |
| | - License (Apache/MIT) | |
| | - Production vs research | |
| | - Create candidate shortlist (3-5 repos) | |
| +------------------------------------------------------------------+ |
| | |
| v |
| PHASE 4: RECONNAISSANCE DOWNLOAD |
| +------------------------------------------------------------------+ |
| | - git clone --depth 1 [repo] ~/reconnaissance/[name] | |
| | - Check size: du -sh | |
| | - Explore structure: ls -la, find *.py | |
| | - Read README.md for architecture | |
| | - Identify core algorithm locations | |
| +------------------------------------------------------------------+ |
| | |
| v |
| PHASE 5: DEEP DIVE EXTRACTION |
| +------------------------------------------------------------------+ |
| | - Read core algorithm files | |
| | - Understand patterns and interfaces | |
| | - Map to project integration points | |
| | - Identify what to KEEP vs DELETE | |
| | - Extract ONLY essential code | |
| +------------------------------------------------------------------+ |
| | |
| v |
| PHASE 6: GUIDE CREATION |
| +------------------------------------------------------------------+ |
| | - Create MANIFEST.md with: | |
| | - Executive summary | |
| | - Algorithm inventory | |
| | - Integration architecture | |
| | - Effort estimates | |
| | - What to use / What to skip | |
| | - Create comparison charts if multiple sources | |
| | - Update project architecture docs | |
| +------------------------------------------------------------------+ |
| | |
| v |
| PHASE 7: CLEANUP |
| +------------------------------------------------------------------+ |
| | - Delete .git/ (largest space consumer) | |
| | - Delete docs/, examples/, tests/ (reference README) | |
| | - Delete algorithms NOT needed (keep only essentials) | |
| | - Target: <200KB per repo | |
| | - Verify: find . -name "*.py" shows only essentials | |
| +------------------------------------------------------------------+ |
| |
+-----------------------------------------------------------------------+
Input Parameters
| Parameter |
Required |
Description |
gap_description |
YES |
What capability is missing |
project_path |
YES |
Path to project with gap |
tech_stack |
YES |
Languages, frameworks, deployment |
priority |
NO |
P0/P1/P2 (default: P1) |
max_repos |
NO |
Max repos to analyze (default: 3) |
Output Artifacts
| Artifact |
Location |
Purpose |
MANIFEST.md |
reconnaissance/[repo]/ |
Integration guide |
*-COMPARISON-CHART.md |
reconnaissance/ |
Multi-source comparison |
*-RESEARCH-*.md |
reconnaissance/ |
Paper analysis |
| Cleaned repo |
reconnaissance/[repo]/ |
Essential code only |
| Architecture update |
Project docs |
Gap now documented |
Tech Stack Compatibility Checklist
languages:
python:
compatible: [Python, Cython]
maybe: [Rust via PyO3, C via ctypes]
incompatible: [Java, Go, pure C++]
rust:
compatible: [Rust]
maybe: [C via FFI, Python via PyO3]
incompatible: [Python-only, JavaScript]
typescript:
compatible: [TypeScript, JavaScript]
maybe: [Rust via WASM]
incompatible: [Python, C++]
deployment:
docker_compose:
compatible: [Docker, docker-compose]
maybe: [Kubernetes with adaptation]
incompatible: [serverless, Lambda-only]
kubernetes:
compatible: [Helm, K8s manifests]
maybe: [Docker with adaptation]
incompatible: [Docker Compose only]
communication:
rest_api:
compatible: [REST, HTTP, gRPC]
maybe: [GraphQL, WebSocket]
incompatible: [proprietary protocols]
Pre-Mortem Checklist
Before committing to a solution, verify:
- [ ] Does project ACTUALLY need this? (check existing code first)
- [ ] Does solution match tech stack?
- [ ] Is solution production-ready or research-only?
- [ ] What's the minimum viable extraction?
- [ ] What can we DELETE after extraction?
- [ ] Are there simpler alternatives?
Example Invocation
User: "fog-compute needs federated learning for training and inference"
Claude executes:
1. GAP: FL training + inference missing (100% gap per architecture doc)
2. CRITERIA: Python, Docker, integrate with NSGA-II scheduler
3. RESEARCH: WebSearch "federated learning framework 2025 github"
4. DOWNLOAD: Clone FATE-LLM, PFLlib
5. EXTRACT: Keep FedKSeed, InferDPT, SCAFFOLD, FedDBE
6. GUIDE: Create MANIFEST.md, comparison chart
7. CLEANUP: 9.5MB -> 126KB, 197MB -> 124KB
Success Metrics
| Metric |
Target |
Rationale |
| Repo size reduction |
>90% |
Keep only essentials |
| Integration effort documented |
<1 week |
Clear roadmap |
| Tech stack match |
100% |
No incompatible solutions |
| Cleanup verified |
<200KB/repo |
Storage efficiency |
Anti-Patterns
| Anti-Pattern |
Why Bad |
Instead |
| Clone entire repo and keep |
Wastes storage, clutters |
Extract only what's needed |
| Skip pre-mortem |
Miss tech stack issues |
Always verify compatibility |
| No MANIFEST.md |
Knowledge lost |
Document integration path |
| Keep .git folder |
80%+ of size |
Always delete .git |
| Research without downloading |
Superficial understanding |
Clone and read code |
Related Skills
literature-synthesis - For paper-heavy researchcodebase-archaeology - For understanding existing codeintegration-architect - For designing integration
Changelog
| Version |
Date |
Changes |
| 1.0.0 |
2026-01-03 |
Initial extraction from FL reconnaissance session |