@appliqation/dashboard

Appliqation Dashboard

Reads the audit trail every agent in the Appliqation agent family writes, and prints an aggregated report: how many runs, how many tokens, and what actually happened, over a period.

Genuinely separate from Appliqation itself — this agent family is a parallel system that uses Appliqation, not a feature of it, so its own audit trail lives in infrastructure this family owns (a MongoDB you configure, or a plain local JSONL file), never inside appq. This tool has no LLM, no appq connection, no credentials beyond pointing at that store.

How it works

Every sibling agent (appliqation-autotest, appliqation-scriptgen, appliqation-defect-fix, appliqation-pr-raise, appliqation-explorer, appliqation-autopilot) optionally writes one record per invocation — token usage, duration, and its own real --json outcome — via @appliqation/agent-core's audit/sink.ts, entirely opt-in and best-effort (a broken audit store never affects the real run). This tool reads the same store back and aggregates it.

appliqation-dashboard report --group-by month

2026-08-01 to 2026-08-31, grouped by month

Period   Agent                    Runs  In tokens  Out tokens  Cache read  Duration  Derived              Est. cost
-------  -----------------------  ----  ---------  ----------  ----------  --------  -------------------  ---------
2026-08  appliqation-autotest     42    850000     120000      600000      95.2s avg 118 test cases judged $4.20
2026-08  appliqation-explorer     6     210000     40000       50000       210.5s avg 5 completed passes   $1.10
2026-08  appliqation-pr-raise     8     0          0           0           3.1s avg   6 PRs opened/updated -

Quick start

npm install -g @appliqation/dashboard

Create a .env file (in whatever directory you'll run it from) pointing at the same audit store your agents write to — see Configuration below.

appliqation-dashboard report [--since <date>] [--until <date>] [--group-by day|week|month|quarter|year] [--agent <name>]

Defaults to the last 30 days, grouped by day, all agents. --group-by accepts day/week/month/quarter/year.

Configuration

Copy .env.example to .env. Point AUDIT_MONGO_URI/AUDIT_MONGO_DB/AUDIT_MONGO_COLLECTION (or AUDIT_JSONL_PATH) at the same store the sibling agents' own .env files write to — this tool errors clearly if neither is configured, rather than silently printing an empty report. MODEL_PRICING optionally points at a small JSON file of per-model $/1M tokens prices to estimate cost; omitted entirely from the report if unset — never a stale hardcoded price.

Development

git clone https://github.com/appliqation/dashboard.git
cd dashboard
npm install
cp .env.example .env   # point at the same audit store your agents write to
npm run dev -- report [--group-by month]
npm run typecheck
npm test

See CLAUDE.md for a map of this repo if you're working in it with an AI coding assistant.

License

MIT — see LICENSE.