The observability console for Claude Code and Codex — local-first, permanent, and honest about cost.
Claude Code and OpenAI Codex write a detailed transcript of every session — every turn, every token, every tool call, every sub-agent they spawned — and then tell you almost nothing about it. Argus tails those files into a SQLite archive on your machine, prices each turn, and serves a dashboard that answers the questions the transcripts never do: what am I spending, where did it go, and which turn made it so?
$ pip install argus-code$ argus start
Argus keeps the structure of every session — turns, tool calls, sub-agents, cache reads — not just its sums. So you can find the turn that blew the budget, the tool that quietly started failing last Tuesday, and exactly what a sub-agent was told before it went off the rails.
Detectors re-check your data every 10 minutes against historical baselines and file alerts — a tool whose error rate doubles, a project whose weekly spend triples, a prompt cache that quietly stopped hitting — all flagged the day it happens, not when you happen to notice.
Claude Code rotates its own logs, so tools that read what is on disk see a sliding month. Once Argus has ingested a session the row stays forever; a few months in, Argus remembers sessions your agents have already forgotten.
A virtualised, sortable grid of every session you have ever run — Claude Code and Codex side by side — with inline token bars and a duration × tokens scatter that makes outliers obvious. Filter by text, agent, project, model or time window; export to CSV.
Click a session and you get its shape — one bar per turn, cache reads beneath fresh tokens, failing turns in red — plus a cumulative-cost line, a tool mix with error segments, and a timeline where every turn expands into its tool calls, sizes, status and error text inline.
When a session delegates to sub-agents, Argus keeps each one: the task as it was given, its tools, tokens, cost, shape and full timeline — with an at-a-glance strip that turns red where an agent failed. No more guessing what that Task or spawn_agent call actually did.
This period vs last, a monthly run-rate projection, tokens per session, cache-read share, and $ per million tokens as you actually experienced it — so you can tell "I am working more" apart from "each session got more expensive".
A leaderboard of every tool with its error share, calls per day stacked by tool, MCP servers, and sub-agent invocations by type. Detectors re-check the last 7 days against a 4-week baseline every 10 minutes and file an alert when a tool's failures double, a project's weekly spend multiplies, or its prompt-cache hit rate falls off a cliff.
Search every prompt you have typed and every assistant reply, thinking block and tool output. SQLite FTS5 — sub-millisecond, lexical, deterministic, offline. Off by default.
Argus is built for one person on one machine, and the defaults say so.
Nothing on your LAN, Wi-Fi or VPN can reach it.
Except argus pricing refresh, a manual command that fetches one JSON file. No telemetry, no analytics, no LLM calls.
Cost and token analytics need no text content. Full-text search over prompts and transcripts requires an explicit opt-in.