add mini-context-graph skill - A persistent, compounding knowledge base

[fizznix]

Pull Request Checklist

• I have read and followed the CONTRIBUTING.md guidelines.
• I have read and followed the Guidance for submissions involving paid services.
• My contribution adds a new instruction, prompt, agent, skill, or workflow file in the correct directory.
• The file follows the required naming convention.
• The content is clearly structured and follows the example format.
• I have tested my instructions, prompt, agent, skill, or workflow with GitHub Copilot.
• I have run npm start and verified that README.md is up to date.
• I am targeting the staged branch for this pull request.

---

Description

Introduces mini-context-graph, a skill that builds persistent, compounding knowledge bases by combining wiki pages with a structured knowledge graph. Unlike standard RAG that re-discovers knowledge on every query, this skill ingests documents once and maintains evolving knowledge that accumulates and cross-references over time.

Architecture

Three-layer knowledge storage:

1. Raw Sources — Original documents stored immutably with chunk metadata
2. Wiki Layer — LLM-written markdown pages with cross-references that grow richer over time
3. Graph Layer — Extracted entities and relations with provenance links back to supporting text

What's included

• SKILL.md — Core concepts, quick start, operations, and constraints
• ingestion.md — Entity/relation extraction rules and workflow
• ontology.md — Type normalization and entity deduplication
• retrieval.md — Query execution and BFS graph traversal
• lint.md — Wiki health checks and maintenance procedures
• scripts/ — Python implementation:
  • contextgraph.py — Main skill interface
  • tools/ — Graph store, wiki store, document store, retrieval engine, and index management

Key features

• Knowledge accumulation — Learns once, answers faster on subsequent queries
• Provenance tracking — Every entity and relation links back to original source text
• Wiki-first retrieval — Fast path using markdown pages before expensive graph traversal
• Evidence-based answers — Returns supporting documents and reasoning chains
• Graph-aware queries — BFS traversal across entities and relations (max depth: 2)
• Automated wiki maintenance — Lint checks for orphaned pages, broken links, and contradictions

---

Type of Contribution

• New instruction file.
• New prompt file.
• New agent file.
• New plugin.
• New skill file.
• New agentic workflow.
• Update to existing instruction, prompt, agent, plugin, skill, or workflow.
• Other (please specify):

---

By submitting this pull request, I confirm that my contribution abides by the Code of Conduct and will be licensed under the MIT License.

[github-actions]

🔍 Skill Validator Results

✅ All checks passed

Scope	Checked
Skills	1
Agents	1
Total	2
Severity	Count
---	---:
❌ Errors	0
⚠️ Warnings	0
ℹ️ Advisories	0

Summary

Level	Finding
ℹ️	Found 1 skill(s)
ℹ️	[mini-context-graph] 📊 mini-context-graph: 1,945 BPE tokens [chars/4: 2,418] (detailed ✓), 25 sections, 2 code blocks
ℹ️	✅ All checks passed (1 skill(s))

Full validator output

```text Found 1 skill(s) [mini-context-graph] 📊 mini-context-graph: 1,945 BPE tokens [chars/4: 2,418] (detailed ✓), 25 sections, 2 code blocks ✅ All checks passed (1 skill(s)) ```

[aaronpowell]

Additional files for a skill should be in the assets or references folder per the Skills spec (and scripts but that's already used in this PR).

[aaronpowell]

Looks like something is messed up on the branch for this PR - might be easier to discard it and start again