docs: community launch — awesome-lists, blog post, marketplace listing

[VECTORG99]

Summary

ACF is invisible to the AI coding community. We have no presence on awesome-claude-skills, awesome-claude-code-and-skills, no blog post, no demo, no benchmark report. Without community presence, ACF won't be adopted regardless of its technical quality. We need a community launch plan.

Context

• Roadmap: docs/ROADMAP.md — Phase 3 (Community)
• Awesome lists found during research:
  • ununununium/awesome-claude-code-and-skills — curated, production-ready skills
  • soummyaanon/awesome-claude-skills — awesome collection with Awesome badge
  • ytrofr/claude-code-guide — 240+ patterns, install tiers
  • MuhammadUsmanGM/claude-code-best-practices — community handbook
• Marketplaces: Agensi (SKILL.md marketplace with install counts)
• Agent Skills spec: agentskills.io — the open standard ACF follows

Motivation

In 30 years of technology, I've seen hundreds of technically superior products die from obscurity while inferior products won because they were visible. The AI tooling market in 2026 is the most crowded I've ever seen:

Agent platforms (6+): Claude Code, Cursor, Codex CLI, OpenCode, Devin, OpenClaw, Aider, Continue, Tabnine, GitHub Copilot, Windsurf, Cline

Skill ecosystems (4+): Agent Skills spec (Anthropic), Cursor rules, OpenCode skills, Agensi marketplace

Awesome lists (10+): awesome-claude-skills, awesome-claude-code-and-skills, awesome-cursor, awesome-ai-agents, awesome-llm-tools

Context management tools (5+): Kimi CLI compaction, Claude Code's built-in context management, Cursor's codebase indexing, Aider's repo map, Continue's codebase context

ACF's differentiators (compaction + caveman + multi-agent + SDLC-specific) are strong, but differentiation without visibility is invisible. The community doesn't know ACF exists.

The launch needs to be honest: ACF is alpha, the benchmarks aren't done yet, and adoption is zero. But the architecture is sound, the compaction research is thorough, and the multi-agent compatibility is real. That's more than most skills can say.

Affected Files

• README.md — add badges (Agent Skills spec, license, CI status)
• docs/ROADMAP.md — update Phase 3 checklist
• docs/BENCHMARKS.md (depends on feat: benchmark compaction and caveman token savings across real projects #2) — needed before launch
• docs/COMPATIBILITY.md — needed before launch (done)

Acceptance Criteria

• README.md updated with:
  • Badges: Agent Skills spec compliant, MIT license, CI status
  • "Supported agents" section with logos/names
  • Quick-start example (install + craft an issue in 3 commands)
  • Comparison table vs other context management approaches
• ACF submitted to awesome-claude-skills (via PR to the awesome list repo)
  • Submission follows the awesome list contribution guidelines
  • Description is concise and highlights differentiators (compaction, caveman, multi-agent)
• ACF submitted to awesome-claude-code-and-skills (via PR)
• Blog post / demo written (can be a GitHub Discussion or README section):
  • "ACF in action" — walk through crafting an issue for a real project
  • Show the compaction tiers (full → compacted → caveman) with token counts
  • Show the multi-agent install (one command, 6 agents)
• Agensi marketplace listing created (if benchmarks are ready from feat: benchmark compaction and caveman token savings across real projects #2)
• ACF added to agentskills.io community skills list (if they maintain one)
• Social post (Mastodon/LinkedIn/Twitter) announcing ACF with link to the repo
• docs/ROADMAP.md Phase 3 checklist updated with submission status

Validation

• ACF appears in at least 1 awesome list (merged PR)
• README has badges that render correctly
• Blog post / demo is readable and accurate (no claims that aren't backed by data)

Complexity

Simple — mostly submissions and writing. But requires benchmarks (#2) and compatibility testing (#4) to be done first for honest claims.

Notes

• Do not submit to awesome lists until benchmarks (feat: benchmark compaction and caveman token savings across real projects #2) and compatibility testing (test: verify ACF installs and triggers correctly on all 6 supported agents #4) are done. Submitting an unvalidated tool damages credibility.
• The blog post should be honest about alpha status — the AI community is skeptical of overhyped tools.
• The comparison table should compare against: Kimi CLI compaction, Cursor's codebase indexing, Aider's repo map, and plain CLAUDE.md/AGENTS.md (no context management).
• This is a Phase 3 (Community) issue — it depends on Phase 1 and Phase 2 being substantially complete.
• Consider reaching out to the Agent Skills spec maintainers (Anthropic) for feedback — they may feature ACF in their example skills.