← Back to Articles & Artefacts
artefactswest

Agent / Plugin Scan — Wave 1

IAIP Research
deep-search--karpathy--llm-wiki--and--rise-framework--qmd-episodic--implication

Agent / Plugin Scan — Wave 1

Highest-leverage patterns already present

ArtifactTypeWhy it matters for issue #195Evidence
context-engineeringpluginGives the kit a clean context-map entrypoint before edits or synthesis./workspace/repos/miadisabelle/mia-awesome-copilot/plugins/context-engineering/README.md:1-27; /workspace/repos/miadisabelle/mia-awesome-copilot/plugins/context-engineering/.github/plugin/plugin.json:1-24
structured-autonomypluginEncodes a plan / generate / implement split that matches Miadi's orchestrator-vs-executor discipline./workspace/repos/miadisabelle/mia-awesome-copilot/plugins/structured-autonomy/README.md:1-20; /workspace/repos/miadisabelle/mia-awesome-copilot/plugins/structured-autonomy/.github/plugin/plugin.json:1-15
doublecheckpluginSupplies a reusable verification layer for promoted wiki claims and literature findings./workspace/repos/miadisabelle/mia-awesome-copilot/plugins/doublecheck/README.md:1-37
context-architectagentStrong model for context-first multi-file planning./workspace/repos/miadisabelle/mia-awesome-copilot/agents/context-architect.agent.md:7-26
repo-architectagentStrong model for scaffolding agent/plugin/skill structures and hybrid CLI/editor environments./workspace/repos/miadisabelle/mia-awesome-copilot/agents/repo-architect.agent.md:8-20,31-51,64-96
task-researcheragentClosest existing "wiki maintainer" behavior: writes evidence-backed markdown research, removes duplicates, prefers one chosen approach./workspace/repos/miadisabelle/mia-awesome-copilot/agents/task-researcher.agent.md:11-37,39-52
create-llms + update-markdown-file-index + create-agentsmdskillsReady-made primitives for bootstrapping llms.txt, indexes, and agent-facing schema files./workspace/repos/miadisabelle/mia-awesome-copilot/skills/create-llms/SKILL.md:6-47,49-80; /workspace/repos/miadisabelle/mia-awesome-copilot/skills/update-markdown-file-index/SKILL.md:6-18,53-77; /workspace/repos/miadisabelle/mia-awesome-copilot/skills/create-agentsmd/SKILL.md:6-24,195-249
remember + memory-mergerskillsUseful model for turning session lessons into durable, domain-scoped markdown memory that later gets merged into stable instructions./workspace/repos/miadisabelle/mia-awesome-copilot/skills/remember/SKILL.md:21-30,69-107; /workspace/repos/miadisabelle/mia-awesome-copilot/skills/memory-merger/SKILL.md:37-90

What this repo teaches the Miadi-native kit

  1. Plugins are the right packaging boundary for reusable capability.
    This repo packages agents/skills declaratively and validates them via manifest/schema discipline (/workspace/repos/miadisabelle/mia-awesome-copilot/AGENTS.md:101-108,138-158,160-179).

  2. Research, planning, and implementation should not be collapsed.
    task-researcher, context-architect, and structured-autonomy all keep discovery and execution separate. That matches Miadi's PDE + subagent orchestration posture and should be copied into the kit.

  3. Verification needs its own reusable surface.
    doublecheck proves that verification can be packaged as a first-class plugin rather than left as an ad hoc habit.

  4. Markdown maintenance is a capability, not just documentation.
    create-llms, update-markdown-file-index, create-agentsmd, remember, and memory-merger show that repo-shaping markdown work already fits the plugin/skill ecosystem cleanly.

Boundary signal: kit vs suite

Best fit for the Miadi-native Copilot orchestration kit

  • context map / dependency trace
  • plan / generate / implement orchestration wrappers
  • wiki-page drafting and merge review
  • repo-local memory capture and promotion
  • claim verification before page promotion

Best fit for the broader Miadi Orchestration Suite

  • shared schemas and manifest validation
  • retrieval/indexing services
  • audit/session logging
  • governance review and policy enforcement
  • cross-runtime plugin/subagent distribution

Immediate borrow list

  1. context-engineering shape for the kit's "read first, map context" action.
  2. structured-autonomy split for orchestrator vs executor responsibilities.
  3. doublecheck posture for wave-2 literature and wiki-promotion verification.
  4. create-llms / update-markdown-file-index / create-agentsmd primitives for wiki scaffolding.
  5. remember / memory-merger pattern for durable lessons and periodic consolidation.