NousResearch
Featured
hermes-agent
Hermes Agent is an open-source, self-improving AI agent with a built-in learning loop that creates skills from experience, runs across platforms, and supports any LLM provider.
Why it is worth attention
It provides honest, privacy-respecting observability for coding agents by joining session usage data with recorded work, labeling each attribution with a confidence level, and never phoning home or requiring an API key.
Who it is for
- developers using Claude Code or Codex
- engineering teams managing agent costs and output
- open-source maintainers seeking agent transparency
- platform engineers evaluating agent accountability
Use cases
- monitor per-agent token consumption and estimated cost
- review tasks with recorded work steps and verification evidence
- audit attribution quality between usage and work logs
- compare agent performance across sessions without cloud dependency
Strengths
- privacy-first: everything stays local on your machine
- honest labeling: costs are clearly marked as estimates, attribution gaps are shown
- supports multiple coding agents with per-capability matrix
- open-source and extensible with documented reference and architecture
Considerations
- early alpha – interfaces may change; limited to Claude Code and Codex for full usage paths
- requires Python ≥3.11 on macOS/Linux; Windows only via WSL
- no cloud sync or hosted dashboard; all state is local files
README quick start
agentacct
See what your coding agents actually did — and what it cost — on a dashboard that never leaves your machine.
agentacct is local-first Agent Work Intelligence for coding agents. It reads the session logs that Claude Code and Codex already write on your machine, joins them with the work each session records as it goes, and shows the result — tokens, estimated cost, tasks, and evidence — on a local dashboard.
Private by design. Everything stays on your machine: state is plain local files, the dashboard binds to 127.0.0.1, and there is no phone-home telemetry, no account, no cloud sync. agentacct never stores or requests a provider API key.
Screenshots show a synthetic demo workspace; your dashboard renders your machine's real local data.
What you get
- Honest usage and cost. Tokens per agent, model, and day — read from the clients' own local session files and labeled
client_reported; costs are clearly marked pricing-table estimates, never invoices. - The work, not just the tokens. Sessions roll up into Tasks with recorded work steps and machine checks: a passing test is
Verifiedevidence, an agent's own claim stays labeledAgent reported. - Attribution you can trust. Every join between usage and recorded work carries a confidence label (
exact/high/medium/low). Missing attribution beats wrong attribution: when agentacct cannot prove a link, it shows the gap instead of a guess.
Install
Requires Python >= 3.11 on macOS or Linux; Windows is supported only via WSL.
pipx install agentacct
agentacct onboard # from your repo root
onboard detects your local coding-agent logs, sets up the project-local store, runs a first usage sync, and starts the dashboard at http://127.0.0.1:8765. Then open a new agent session in the project — MCP servers and hooks bind at session start, so the session that ran onboarding cannot become the first recorded Task.
Prefer to let the agent do it? Paste this into the coding agent working in your target repo:
Install and set up agentacct in this repo — a local-first dashboard that reads my
coding-agent logs read-only and shows honest token usage and cost.
Run `pipx install agentacct`
(or `pipx install git+https://github.com/mikehasa/agentacct`),
then `agentacct onboard` from this repo root, then tell me the dashboard URL.
Observe-only: never
Description
Local-first Agent Work Intelligence for coding agents: usage truth, recorded work, and honest joins. Read-only over coding-agent logs; zero-JavaScript localhost dashboard.
Related repositories
Similar projects matched by category, topics, and programming language.
lopopolo
Featured
harness-engineering
Harness Engineering is a methodology for improving coding agent outputs by carefully crafting the environment around them—providing curated context, tools, and executable constraints that encode an organization’s nonfunctional requirements and cumulative lessons.
littledivy
Featured
mimic
mimic captures traffic from any iOS or web app and automatically generates a Python client library that lets you call the app's API like a regular library.