atsuki-shirasawa/kairos
A personal web app for looking back on your Claude Code session history as a calendar of what you did
ARCHETYPE
Orchestrator
A bench of specialised subagents. The main agent mostly delegates.
CONTEXT TAX · EVERY TURN
up to ~4.1k tokens
Claude Code loads MCP tools on demand, so the real cost is usually lower.
Moderate · median rig: 2.3k · breakdown
Copy this rig
# review before running: this installs third-party code $ claude mcp add context7 -- npx -y @upstash/context7-mcp $ npx degit atsuki-shirasawa/kairos/.claude ./rig-kairos # inspect, then merge into .claude/
MCP servers are added to Claude Code at local scope; env vars are shown as YOUR_… placeholders — we never store values. Files are fetched with degit into a separate folder so you can review before merging.
$ claude mcp add context7 -- npx -y @upstash/context7-mcp $ npx degit atsuki-shirasawa/kairos/.claude/skills/add-api-route .claude/skills/add-api-route $ npx degit atsuki-shirasawa/kairos/.claude/skills/add-fixture-scenario .claude/skills/add-fixture-scenario $ npx degit atsuki-shirasawa/kairos/.claude/skills/add-migration .claude/skills/add-migration $ npx degit atsuki-shirasawa/kairos/.claude/skills/add-model-price .claude/skills/add-model-price $ npx degit atsuki-shirasawa/kairos/.claude/skills/land .claude/skills/land $ npx degit atsuki-shirasawa/kairos/.claude/skills/ship-local .claude/skills/ship-local $ npx degit atsuki-shirasawa/kairos/.claude/skills/verify-ui .claude/skills/verify-ui
$ curl -fsSL --create-dirs -o .claude/agents/design-reviewer.md https://raw.githubusercontent.com/atsuki-shirasawa/kairos/main/.claude/agents/design-reviewer.md $ curl -fsSL --create-dirs -o .claude/agents/i18n-reviewer.md https://raw.githubusercontent.com/atsuki-shirasawa/kairos/main/.claude/agents/i18n-reviewer.md $ curl -fsSL --create-dirs -o .claude/agents/log-format-auditor.md https://raw.githubusercontent.com/atsuki-shirasawa/kairos/main/.claude/agents/log-format-auditor.md $ curl -fsSL --create-dirs -o .claude/agents/query-plan-reviewer.md https://raw.githubusercontent.com/atsuki-shirasawa/kairos/main/.claude/agents/query-plan-reviewer.md $ curl -fsSL --create-dirs -o .claude/agents/security-reviewer.md https://raw.githubusercontent.com/atsuki-shirasawa/kairos/main/.claude/agents/security-reviewer.md $ curl -fsSL --create-dirs -o .claude/agents/structure-reviewer.md https://raw.githubusercontent.com/atsuki-shirasawa/kairos/main/.claude/agents/structure-reviewer.md
Merge into .claude/settings.json (project) or ~/.claude/settings.json (user). Hook commands reference scripts in the source repo — copy those too.
{
"hooks": {
"PreToolUse": [
{
"matcher": "Edit|Write|MultiEdit",
"hooks": [
{
"type": "command",
"command": "\"$CLAUDE_PROJECT_DIR\"/.claude/hooks/protect.sh"
}
]
}
]
}
} MCP servers (1)
| server | source | est. tokens |
|---|---|---|
| Context7 | npm | 1.2k |
Skills (7)
Subagents (6)
| design-reviewer model: sonnet | Reviews UI changes in src/web against DESIGN.md — theme tokens instead of hard-coded colors, the single indigo accent, type sizes and tones, block fills and states, and DESIGN.md / index.css staying i |
| i18n-reviewer model: sonnet | Reviews UI changes in src/web for i18n mistakes that the type checker can't catch — hard-coded copy in components, message functions called at module level, and Japanese that drifts from the English s |
| log-format-auditor model: sonnet | Scans real Claude Code logs (~/.claude/projects) read-only and reports drift from Kairos's interpretation rules (the rules table in tests/fixtures/README.md), as well as unknown record types and field |
| query-plan-reviewer model: sonnet | Checks the SQL that a diff adds or changes in src/server/queries and the other modules that run SQL (the summarizer's store and ledger, the live reads, ingest statements), and schema changes in src/se |
| security-reviewer model: opus | Security review for Kairos. Assuming the local server serves conversation logs from every project, checks DNS rebinding, CSRF, XSS, leaks to external services, and side effects of running claude -p. U |
| structure-reviewer model: opus | Reviews a diff in src/ against .claude/rules/code-structure.md — single responsibility, the signals to split, where logic lives per layer, and the problems this codebase has had (god façades, DTOs map |
Hooks (3)
| event | matcher | runs |
|---|---|---|
| PreToolUse | Edit|Write|MultiEdit | "$CLAUDE_PROJECT_DIR"/.claude/hooks/protect.sh |
| PostToolUse | Edit|Write|MultiEdit | "$CLAUDE_PROJECT_DIR"/.claude/hooks/format.sh |
| Stop | * | "$CLAUDE_PROJECT_DIR"/.claude/hooks/rules-check.sh |
Permissions
deny (0)
—
ask (0)
—
allow (1)
Bash(sqlite3 -readonly:*)
Similar rigs
NathanAB/statuslin.es
Community gallery of Claude Code status lines.
Automator 6.5k tok · · Q32
codeverbojan/claude-code-kickstart
Production-grade agentic workflow template for Claude Code. 4-layer architecture, task playbooks, session memory, sub-agents, one-click install.
Orchestrator 10.2k tok · · Q15
BhodiSea/next-expo-supabase-agent-harness
A CLI that scaffolds a Next.js and Expo (React Native) monorepo on Supabase, with Claude Code hooks and CI gates that block an agent turn or a merge until validation passes.
YOLO Cowboy 11.1k tok · · Q54
cogvera-labs/next-expo-supabase-agent-harness
A CLI that scaffolds a Next.js and Expo monorepo on Supabase, with Claude Code hooks that send the agent back to work when a check fails.
YOLO Cowboy 11.1k tok · · Q54