--- okf_version: "0.1" bundle: aria-coding-agent-ecosystem source_bundle: "coding-agent-ecosystem" title: "OKF Knowledge Bundle — Coding-agent ecosystem" updated: 2026-07-06 concepts: 15 --- # Coding-agent ecosystem — OKF v0.1 Knowledge Bundle Claude Code skills, plugins, orchestration frameworks and spec-driven workflows from public GitHub projects and community walkthroughs. Self-contained, public-source AI-engineering material — no employer or personal data. Each entry is one concept (one markdown file); the `type` tag mirrors its role in the source encyclopedia. --- ## Concepts (15) - [agent-skills-engineering-workflow-osmani](./agent-skills-engineering-workflow-osmani.md) — Addy Osmani's agent-skills: 24 portable Markdown skills + 7 slash commands mapping senior-engineering discipline onto the SDLC for AI coding agents. - [awesome-claude-code-subagents](./awesome-claude-code-subagents.md) — A large MIT-licensed catalog of 154+ ready-made Claude Code subagent definitions across 10 categories to drop into .claude/agents/ as a starter roster. - [claude-code-plugins-roundup-chase](./claude-code-plugins-roundup-chase.md) — A Chase AI walkthrough of 17 add-ons that extend Claude Code, grouped Design / Productivity / Data, spanning agent skills, plugins, and CLIs. - [claude-code-spec-driven-development](./claude-code-spec-driven-development.md) — Claude Code already ships four isolated state layers — plan mode, a committed spec, a live task list, and a durable todos.json mirror — keep each doing one job. - [claude-design-claude-code-integration](./claude-design-claude-code-integration.md) — Anthropic's Claude Design canvas and its two-way Claude Code bridge: /design-sync of repo tokens/components, a canvas-to-terminal handoff, a harness-injected MCP. - [claude-website-seven-levels-design-skills](./claude-website-seven-levels-design-skills.md) — Jack Roberts' seven-level ladder for building beautiful websites with Claude Code (Fable 5): grab-and-go → references → design skills (GitHub) → image/video gen → UI sniping → data → design extraction. - [design-skills-for-coding-agents](./design-skills-for-coding-agents.md) — A tour of SKILL.md design skills for coding agents (direction, shadcn, GSAP, taste presets, mobile) that hand the model a design decision before writing code. - [gas-town-agent-orchestrator](./gas-town-agent-orchestrator.md) — Steve Yegge's Go orchestrator runs many Claude Code workers across projects on a Git-backed work queue (Beads) so long workflows finish despite agent crashes. - [gstack-claude-code-engineering-team](./gstack-claude-code-engineering-team.md) — gstack: an open-source bundle of ~23 opinionated Claude Code role-skills, turning one session into a Think-Plan-Build-Review-Test-Ship-Reflect sprint pipeline. - [oh-my-claudecode-orchestration](./oh-my-claudecode-orchestration.md) — OMC is an MIT plugin+CLI wrapping Claude Code in a staged plan→prd→exec→verify→fix pipeline driving 19 specialised sub-agents with cost-aware model routing. - [open-source-ai-projects-roundup-berman](./open-source-ai-projects-roundup-berman.md) — Matthew Berman curates 12 open-source AI repos — agent skills, MCP servers, and Claude-Code-ecosystem tooling — each verified against its own README and GitHub API. - [pixel-agents-multiagent-visualizer](./pixel-agents-multiagent-visualizer.md) — An open-source VS Code extension renders Claude Code agent activity as animated pixel-art characters, inferred purely by watching JSONL transcripts. - [ponytail-laziness-ladder-skill](./ponytail-laziness-ladder-skill.md) — A cross-agent coding skill that stops an AI agent at the first rung of a YAGNI→stdlib→native→one-line ladder, cutting ~54% of lines while keeping safety guards intact. - [skill-ecosystem-three-artifacts-berman](./skill-ecosystem-three-artifacts-berman.md) — A Matthew Berman walkthrough of three live Agent Skills ecosystem artifacts: a capability skill, a workflow skill bundle, and a context-compression layer. - [vercel-eve-agent-framework](./vercel-eve-agent-framework.md) — Vercel's eve defines an agent as a directory of files — instructions, tools, skills, channels — compiled into an app running on Vercel's serverless substrate. --- --- theme: agents type: meta status: synthesized updated: 2026-07-05 source_bundle: "coding-agent-ecosystem" --- # Coding-agent ecosystem — OKF v0.1 Knowledge Bundle ## What this is A self-contained, shareable **OKF v0.1 Knowledge Bundle** promoted from the ARIA-RD encyclopedia's **`coding-agent-ecosystem`** logical bundle: Claude Code skills, plugins, orchestration frameworks and spec-driven workflows from public GitHub projects and community walkthroughs. It gathers 14 public-source AI-engineering concept pages plus an index and a change log. Everything here is built from **public / external sources** — there is **no employer-specific or personal content**. ## Layout ``` index.md catalog / entry point (source_bundle declared in frontmatter) README.md this file log.md change log okf-cli.py dependency-free consumer (stdlib; BM25 + optional semantic search) *.md 15 concept pages (flat, bundle root) ``` Each concept is one markdown file whose id is its filename minus `.md`. Every non-reserved `.md` carries a `type` field in its frontmatter — the OKF conformance contract. `index.md` and `log.md` are reserved (excluded from search). ## How to use ```bash python okf-cli.py list # concept ids python okf-cli.py index # the annotated catalog python okf-cli.py search "your query" # BM25-ranked search with snippets python okf-cli.py search "your query" --semantic # embeddings + cosine (needs Ollama) python okf-cli.py read # print a concept python okf-cli.py selftest # OKF-conformance check ``` `search` ranks with BM25 over titles/descriptions/bodies. `--semantic` uses the precomputed `embeddings.json` (if present) and a local Ollama for the query vector, degrading gracefully to BM25 when either is unavailable. The legacy `find` keyword command is retained for backward compatibility. Cross-links between the 14 concepts are bundle-relative (`./slug.md`), so the bundle is navigable offline. ## External references These pages live in the full ARIA-RD encyclopedia (not in this bundle); inline links to them were unwrapped to plain text: - `26-01-07 gaz town` - `agent-economy` - `agent-long-term-memory-solutions` - `agent-skills-progressive-disclosure` - `anthropic-agent-sdk-billing-management` - `claude-code-observability-command-center` - `claude-fable5-youtube-content-pipeline` - `context-engineering` - `deepresearch-agent-orchestration` - `github-repos-to-keep` - `hermes-agent` - `hermes-agent-mcp-server-claude-code` - `llm-council-multi-model-deliberation` - `mcp-and-a2a-interop` - `omnigent-meta-harness` - `pai-life-operating-system` - `pencil-dev-ide-design-canvas` - `rag-pipeline-architecture` - `remote-ai-agent-setup` - `selling-to-ai-agents-agentic-web` - `skill-chaining-agent-composition` - `spec-driven-ai-development` - `tailwind-shadcn-ui-stack` - `temporal-knowledge-graphs-graphiti` - `understand-anything-codebase-knowledge-graphs` - `voicebox-local-voice-studio` ## Provenance & interface Promoted from the **ARIA-RD** R&D AI-knowledge encyclopedia on **2026-07-05** from the `coding-agent-ecosystem` logical bundle. `source_bundle: "coding-agent-ecosystem"` (in this README and `index.md`) is the **interface contract** a nightly sync reads to keep this shipped bundle aligned with the logical bundle's live membership. This bundle is a copy — the encyclopedia remains the source of truth. See `log.md` for history.