[opensourcedex]

agentacct is a free, open source frameworks & platforms project written in Python and released under MIT. It has 757 GitHub stars, 80 forks and 60 open issues, and was last pushed 1 months ago. On this registry it ranks #127 of 139 tracked projects in Frameworks & Platforms, with 5 head-to-head comparisons available. It gained 1 stars over the last 5 tracked days.

What is agentacct?

agentacct is a local-first, MIT-licensed Python dashboard and terminal app that shows developers and teams what their coding agents did and what it cost, breaking each task into work steps such as tools used, files changed, tests run, and tokens spent.

What it is

agentacct is a Python package published on PyPI and run as a command-line tool, offering both a terminal interface opened with "agentacct tui" and a signed, notarized macOS app downloadable from the latest release. It lives in the AI coding agent tooling ecosystem: it reads the local session logs of Claude Code, Codex, Kimi Code, OpenCode, Hermes, and others, and turns them into sessions, Work Receipts, work groups, a dashboard, and a usage view. The license is MIT, the repository has 756 stars and 80 forks, and its topics include agent-observability, cost-tracking, claude-code, codex, and llm. Python 3.11 or newer is required on macOS or Linux, and Windows is supported only through WSL.

The concrete problem is that an agent's own account of its work is scattered across separate per-agent session logs and is taken at face value. agentacct replaces that with a single latest-first task list in which an agent's own "done" is marked Reported, while a Verified result requires current passing checks after the last recorded change. Claims and checks stay separate, a failed attempt stays visible in the history alongside later results, and estimated cost and token spend are attached to the task rather than guessed at afterwards. Nothing is sent to an account or a cloud service; the data stays on the machine that produced it.

Key capabilities

  • A Work Receipt per task that gathers the participating sessions, the recorded outcome, estimated cost, claims, check results, and an activity timeline, with steps and full check details one click away.
  • A Reported versus Verified distinction, where verification depends on current passing checks after the last recorded change, and claims remain separate from check results.
  • Work groups that are pointed at a project folder to show which agents worked there, with a shared timeline and drilldown, and group totals computed as sums of the participating sessions with partial coverage and overlap labeled.
  • A Dashboard of the latest recorded work across projects, showing client, activity time, outcome, evidence, and cost, with historical failures and blockers folded into a Recorded issues list.
  • A right-click action on a Recorded issues row that copies a brief assembled from recorded facts.
  • A Usage view with independent Date, Agent, Model, and Provider filters, range totals by agent and by model, a daily chart with a Fresh/All tokens toggle, and per-date breakdowns with cache reads and cache writes reported separately.
  • Ingestion from the local session logs of Claude Code, Codex, Kimi Code, OpenCode, Hermes, and others, with no login and no telemetry.

Who uses it and how

  • Individual developers who run several coding agents on one machine and want a single latest-first list of tasks rather than checking each agent separately.
  • Teams that point a work group at a project folder to see which agents worked there, what they did most recently, and how the sessions add up.
  • Reviewers who need evidence rather than an agent's assertion, using check results and the Reported versus Verified split, and copying a brief from Recorded issues when a failure needs context.
  • Cost-conscious users tracking estimated spend and tokens across a day or a range, filtered by agent, model, and provider, with cache reads and writes kept distinct.
  • Privacy-constrained environments, since there is no account, no cloud component, and nothing leaves the computer; macOS users without a Python install can use the signed app.

Getting started

Install with "pipx install agentacct", then run "agentacct onboard" once per machine to find your agents, set up a local store, and start the recorder, followed by "agentacct tui" for the terminal app. Alternatively, download the signed and notarized macOS app, which requires macOS 14 or later; either way, open a new agent session afterwards because hooks and MCP servers bind at session start.

How it compares

The facts provided do not name any comparable observability product, so agentacct stands alone in this registry rather than displacing a named alternative. It should be read as a companion to the agents whose local logs it consumes, not as a replacement for them.

When to use it — and when not to

Choose it when the work happens on a single machine or a shared project folder and the priority is a private, self-contained record of what agents did and what they cost. Avoid it if you need a hosted, centrally aggregated view across many machines, since the facts describe no account or cloud component, or if you are on native Windows, which is supported only through WSL, or on macOS older than 14 for the app. The project also carries 60 open issues against 756 stars, so expect a tracker with substantial open work rather than a settled, quiet codebase.

project readme (upstream, from github) — read inline

agentacct

English · 简体中文

tests PyPI Python License: MIT

agentacct answers one simple question: what the fuck are my agents actually doing?

See what your coding agents have been working on, across projects and clients: recent activity, tool calls, recorded steps, checks, tokens, and estimated cost. Open a task for its Work Receipt and follow the activity timeline from attempt to result. Built from the local session logs of Claude Code, Codex, Kimi Code, OpenCode, Hermes, and others. No account, no cloud, nothing leaves your computer.

Sessions and a Work Receipt: a latest-first task list beside the recorded outcome, session count, estimated cost, claim coverage, check results, and activity timeline.

Screenshots show a synthetic demo workspace; your install renders your own local data.

Install

pipx install agentacct
agentacct onboard   # once per machine: finds your agents, sets up a local store, starts the recorder
agentacct tui       # the terminal app

Requires Python >= 3.11 on macOS or Linux; Windows is supported only via WSL. For a native window with no Python, download the signed, notarized macOS app from the latest release (macOS 14+).

Either way, open a new agent session afterwards: hooks and MCP servers bind at session start.

The per-agent setup, a --scope project install, and the uv/venv alternatives are in INSTALL.md.

Follow what happened

Sessions opens with the most recently active tasks. Search by task or project, or choose another sort when you need it. Each receipt brings together the participating sessions, recorded outcome, estimated cost, claims, and check results; its activity timeline comes first, with steps and full check details one click away.

An agent's own "done" is Reported. A Verified result needs current passing checks after the last recorded change. Claims and checks stay separate, and a failed attempt stays visible in the history alongside later results. A recorded failure is useful context, not automatically a request for you to intervene.

A day's work, across every agent you run

Work groups show participating clients, recent sessions, token totals, and estimated cost, with a drilldown into shared project activity.

Point a work group at a project folder to see which agents worked there and what they did most recently. Open the group for session details and a shared timeline. Each session keeps its own evidence; group totals are sums of the participating sessions, with partial coverage and overlap labeled.

Start with recent activity

Dashboard overview with recent session status, recorded task count, seven-day usage, and a latest-activity table. Recorded issues are folded below the activity list.

The Dashboard shows the latest recorded work across your projects, with its client, activity time, outcome, evidence, and cost. Historical failed checks and blockers remain available in Recorded issues. They do not decide the default order or become a list of things you must fix. Right-click a Recorded issues row to copy a brief from recorded facts.

Understand the usage behind the work

Usage with independent Date, Agent, Model, and Provider filters, range totals by agent and by model, a daily chart with a Fresh/All tokens toggle, and a selected date's breakdown by client and model with separate cache reads and writes.

Filter by date, agent, model, or provider—independently, so "Codex on Opus, last 7 days" is one selection instead of a page you reset. This range totals the same rows by agent and by model, and selecting a date narrows the client/model breakdown under the date table. The chart remembers your Fresh or All tokens choice, and the tables show fresh, cache, and total tokens side by side with cache reads and writes separated. Hover a Fresh value for input and output counts.

The top bar always says how fresh the local data is (Local data · just now, or the last saved copy). Dates show session totals by activity date; sessions spanning several days are not split into exact daily consumption. Provider-reported quota windows and reset times are available on the same page. Tokens come from client records; costs are pricing-table estimates marked ≈, never an invoice.

Also in the terminal

agentacct tui: a Needs review block whose first item is a blocked claude-code task with its recorded reason, next step, and MCP-record provenance; a Right now rail with Working now, Capacity, Usage change, and Evidence trust; a Recent work table with outcome, evidence, and estimated cost; and a usage history sparkline.

agentacct tui provides receipts, usage, capacity, and its own review overview in your shell — including the folder-anchored Work tab, where each group draws its sessions across every agent on one cross-agent timeline you can open (↵) and zoom/scrub by keyboard. Its top bar carries the same local-data stamp as the app. Press ? for the keys.

Honest by design

agentacct is early alpha and would rather show you a gap than a guess.

  • Reported is not Verified. An agent's claim can never dress up as verification.
  • Estimates are labeled. ≈$ is a pricing-table estimate, ~$ a known-partial subtotal; there is no invoice access. docs/usage-truth-table.md says what each path can prove.
  • Missing beats wrong. Every join between usage and recorded work carries a confidence label (exact, high, medium, low); an unproven link shows as a gap, not a zero.
  • Support is per capability, not per logo. Claude Code and Codex carry the fullest receipts today; the coverage matrix rates every lane for OpenCode, Hermes, DeepSeek Harness, Kimi Code, OpenClaw, and Cursor separately, and agentacct capabilities agents prints the same matrix from your installed version.
  • It records the work, not the conversation. Tool names and categories, touched files, credential-scrubbed commands, exit codes, tokens, and recorded steps. Never your prompts, the model's responses, or transcripts; see the privacy threat model.
  • Nothing phones home. It reads only the local session files of detected clients, listens only on 127.0.0.1, and never stores or requests a provider API key.

How it works

agentacct is local-first Agent Work Intelligence: two evidence streams, kept separate and joined on real client ids instead of guesses.

  • What it used comes from each client's own session files: tokens as the client reported them, costs as pricing-table estimates.
  • What it did comes from the steps and checks the agent records over MCP as it works, plus machine checks such as test runs. How a check was observed decides its evidence tier, never the agent's wording.
  • The join links the two by session and transcript id and labels every attribution with a confidence. Claude Code binds real ids through an installed hook bridge; Codex and OpenCode are matched from their own session logs at import time; Hermes joins recorded work only on ids the agent passes explicitly. The receipt says which path it used.

Join mechanics, the confidence glossary, and the MCP tool list are in docs/reference.md; per-agent integration detail is in docs/coding-agent-integrations.md.

Let your coding agent install it

Paste this into your coding agent:

Install and set up agentacct — a local-first agent work ledger that reads my
coding-agent logs read-only and shows honest token usage, cost, and recorded work.

Run `pipx install agentacct`
(or `pipx install git+https://github.com/mikehasa/agentacct`),
then `agentacct onboard` (installs once per machine, global by default, zero
files written into the repo), then tell me how to open `agentacct tui` and the
local JSON API at http://127.0.0.1:8765.

Observe-only: never store, request, or echo any API key; all state stays local
on this machine. Don't modify my global client config without showing the exact
command first.

The agent then follows INSTALL.md. agentacct setup prompt --agent prints the same prompt.

Uninstall

agentacct stop                 # stop the managed sync + local API (owned processes only)
agentacct uninstall-autostart  # only if you installed autostart
pipx uninstall agentacct

Then delete the store at ~/.local/state/agentacct/state (keep it if you want the history) and remove the entries that agentacct onboard added to your client configs; they are listed per client in docs/coding-agent-integrations.md. A --scope project install keeps everything in that repo's .agent-sentinel/ directory instead.

More

Frequently asked questions

Is agentacct free to use?

agentacct is open source under the MIT licence. There is no licence fee and no seat count — you can self-host it or, where the project offers one, pay a vendor for a managed version instead.

What does agentacct do?

See what your coding agents did and what it cost. Breaks each task down into work steps — tools used, files changed, tests run, time and tokens spent. Local-fir

What is agentacct written in?

agentacct is primarily written in Python. Its source is publicly available at https://github.com/mikehasa/agentacct, and it has 757 GitHub stars.