Backlog item §C · C5 — mirrored from docs/roadmap/improvements-backlog.md.
What the audit found
Root AGENTS.md +
CLAUDE.md (added with this doc set); mirror the skill into .claude/skills/
or symlink so Claude Code auto-discovers; ship .mcp.json example + per-
client snippets (Claude Code, Cursor, Codex TOML — replacing the site's
hardcoded ssh hermes-vm config, apps/web/components/docs-markdown.ts:57-64);
flesh out agents/openai.yaml beyond its 4-line stub or remove it.
Milestone: Phase 4 — Agent ecosystem depth
Before changing behaviour here, read docs/roadmap/what-is-excellent.md — the contracts listed there must survive the fix — and AGENTS.md for build, test, and review conventions.
Definition of done: the change ships with the test named above (or an equivalent), CI is green, and the corresponding line in the backlog file is checked off in the same PR.
What the audit found
Root
AGENTS.md+CLAUDE.md(added with this doc set); mirror the skill into.claude/skills/or symlink so Claude Code auto-discovers; ship
.mcp.jsonexample + per-client snippets (Claude Code, Cursor, Codex TOML — replacing the site's
hardcoded
ssh hermes-vmconfig,apps/web/components/docs-markdown.ts:57-64);flesh out
agents/openai.yamlbeyond its 4-line stub or remove it.Milestone: Phase 4 — Agent ecosystem depth
Before changing behaviour here, read
docs/roadmap/what-is-excellent.md— the contracts listed there must survive the fix — andAGENTS.mdfor build, test, and review conventions.Definition of done: the change ships with the test named above (or an equivalent), CI is green, and the corresponding line in the backlog file is checked off in the same PR.