AIGent
AGENTS.md2 starsChanged 2 months agoMITView on GitHub
AIGent is a local agent workspace: an Electron desktop client, a Python/FastAPI web server and a direct Telegram bot connection. It uses the DeepSeek API, the official local Codex App Server and an independently authenticated official Claude CLI. The target interaction style is Codex: chats, streaming, tools, plans, reviewed changes, project files, Git and terminal output. viewer3d.js is an ES module. node --check <file> silently exits 0 on a file containing import (Node's module detection swallows the parse error), so…
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 32,671 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository3 other files it gives its agents.
- aigent-farmSkill
- aigent-selfdevSkill
- youtube-uploadSkill
The file
We split this AGENTS.md into the rules it gives an agent
5 rules. About 0 are common, 3 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- could be automatic: a setting could enforce it
- Very specific: line 10
Keep runtime state, caches, generated builds and private test evidence in .local/.
- Distinctive: line 13
Use actual provider usage fields.
- Very specific: line 15
Read README.md for the overview, docs/API.md for the connector, SECURITY.md for boundaries and TODO.md for remaining work.
- Distinctive: line 16
Follow installed official CLI protocols.
- Distinctive: line 17
Verify changes with relevant checks and the visible flow before claiming they work.
≈ “Common” is an estimate: at least 1 in 5 of the 41,581 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
AIGent project instructions
Purpose
AIGent is a local agent workspace: an Electron desktop client, a Python/FastAPI web server and a direct Telegram bot connection. It uses the DeepSeek API, the official local Codex App Server and an independently authenticated official Claude CLI. The target interaction style is Codex: chats, streaming, tools, plans, reviewed changes, project files, Git and terminal output.
Working rules
- Для временных файлов и рабочих файлов используй только папки внутри этого репозитория или выбранной папки проекта. Не создавай рабочие файлы в корне диска пользователя.
- Keep runtime state, caches, generated builds and private test evidence in
.local/. Never commit credentials, browser identities, auth profiles, conversations or user attachments. - Normal project exploration must not inspect private server configuration, account credentials or other sessions. An agent may work in its own assigned workspace, including a workspace inside
.local/workspaces/. - Preserve user files and active sessions. Deleting a chat must not delete project files.
- Use actual provider usage fields. Distinguish cache reads, input/output tokens, API cost estimates and subscription limits. Missing metrics are unknown, not zero.
- Group routine tool events. Keep detailed events available for inspection while showing meaningful progress, approvals, errors and the final result.
- Read
README.mdfor the overview,docs/API.mdfor the connector,SECURITY.mdfor boundaries andTODO.mdfor remaining work. For a short orientation question, inspect these primary files selectively and answer concisely with source filenames. - Follow installed official CLI protocols. Do not extract browser cookies, fabricate login status, bypass provider controls or offer a Claude.ai OAuth/subscription-limit integration through the Agent SDK without provider approval.
- Verify changes with relevant checks and the visible flow before claiming they work.
Checks
python -m pytest
python -m ruff check connector tests run.py
node --check connector/static/app.js
node --check connector/static/desktop-ui.js
node --check connector/static/media-ui.js
node --input-type=module --check < connector/static/viewer3d.jsviewer3d.js is an ES module. node --check <file> silently exits 0 on a file containing import
(Node's module detection swallows the parse error), so an ES module must be checked by feeding it to
node --input-type=module --check on stdin. tests/test_models.py runs that check when node is
installed. The vendored three.js under connector/static/vendor/three/ is upstream code: check it,
never reformat it.
Build the desktop bundle with scripts/build-desktop.ps1. Keep outputs in the project and exclude private runtime state from releases.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

