accenture/mercury-composable/sync-adapters

⚠️ Tool-managed skill provided by agent-memory (provenance: agent-memory-builtin). Don't edit it

in place — fork under a new name, or upstream a fix to the agent-memory project (see SKILLS.md).

This skill is the sync skill adapters operation. It is a runnable script — do not hunt for an npm command, MCP tool, or hand-write the adapter files. Run one of these from the repo root (output is byte-identical across all three; bash needs no runtime install — prefer it, especially in a non-Node project):

bash agent-skills/sync-adapters/scripts/sync-adapters.sh
# no bash? (e.g. native Windows) — use whichever runtime you have:
node agent-skills/sync-adapters/scripts/sync-adapters.mjs
python3 agent-skills/sync-adapters/scripts/sync-adapters.py

For each agent-skills/<name>/SKILL.md it (re)writes all six vendor adapters — .claude/skills/<name>/SKILL.md, .gemini/commands/<name>.toml, .cursor/rules/<name>.mdc, .kiro/skills/<name>/SKILL.md, .github/skills/<name>/SKILL.md, .agents/skills/<name>/SKILL.md — and prunes orphaned adapters it previously generated (identified by their pointer signature; never touches hand-authored vendor files). The .agents/skills/ adapter is the Agent Skills standard dir read by Google Antigravity (agy, the Gemini CLI successor) — Antigravity ignores .gemini/commands, so this is what makes /<name> work there.

It is idempotent and writes only the gitignored adapter dirs — never agent-skills/, never a committed file. Pass --dry-run to preview. The neutral agent-skills/<name>/SKILL.md is always the source of truth; the adapters are thin regenerated pointers (do not commit them).

After syncing, reload your runtime if it loads adapters at startup (e.g. GitHub Copilot CLI: /restart or a skills rescan) for a new /<name> to become available this session.