claude-agent-desktop
AGENTS.md211 starsChanged 12 months agoMITView on GitHub
Electron desktop application for agentic chat with the Claude Agent SDK.
- Reads credentials
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 33,315 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 repository6 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- docxSkill
- frontend-designSkill
- pdfSkill
- workspace-toolsSkill
- xlsxSkill
The file
We split this AGENTS.md into the rules it gives an agent
5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 39
Create .claude/skills/<skill-name>/SKILL.md with name + description.
- Very specific: line 40
Add TypeScript tools under .claude/skills/<skill-name>/scripts/.
- Very specific: line 41
Run bun run dev (or rerun scripts/buildSkills.js) to compile binaries into out/.claude/skills/.
- Distinctive: line 63
After a series of code changes, always run: lint, typecheck, test, and format commands to ensure the code is working as expected.
- Distinctive: line 64
When drafting commit messages, always follow Conventional Commits format.
≈ “Common” is an estimate: at least 1 in 5 of the 42,283 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
Claude Agent Desktop
Electron desktop application for agentic chat with the Claude Agent SDK.
Technology
- Framework: Electron 39 (main / preload / renderer)
- Build: Vite 7 via electron-vite
- UI: React 19 + TypeScript, Tailwind CSS 4
- Runtime: Bun (package manager, scripts, tests)
Layout
src/
├── main/ # Electron main process (app lifecycle + IPC handlers)
├── preload/ # Context bridge exposing safe IPC APIs
├── renderer/ # React UI
└── shared/ # Types shared between processes
.claude/ # Claude Agent SDK skills compiled into the app bundle
resources/ # Bundled runtime binaries (bun, uv, etc.)
scripts/ # Build hooks (preDev, beforeBuild, afterPack, runtime downloads)
static/ # Icons and static assetsSkills integration
- Skills live in
.claude/skills/<skill-name>/(each containsSKILL.md+scripts/). - Built-in sample:
workspace-toolswith a depth-limitedlist-directoryutility. scripts/buildSkills.jscompiles TypeScript tools in.claude/skillstoout/.claude/skills/using the root dependencies (no separate.claudepackage) and Bun--compile.- Skill TypeScript is checked via the root
tsconfig.json; there is no package.json or node_modules under.claude. scripts/preDev.jsruns beforebun run devto download runtime binaries and build skills.scripts/beforeBuild.jsruns during production builds to download binaries, copy runtime deps toout/node_modules, and build skills.scripts/afterPack.jstrims unused vendor assets and confirms.claude/skillsare present.- On launch,
src/main/lib/config.tssyncs the bundled.claudedirectory into the workspace so skills are available at runtime.
Adding a skill
- Create
.claude/skills/<skill-name>/SKILL.mdwithname+description. - Add TypeScript tools under
.claude/skills/<skill-name>/scripts/. - Run
bun run dev(or rerunscripts/buildSkills.js) to compile binaries intoout/.claude/skills/. - Launch the app; the workspace
.claudefolder will be refreshed automatically.
Commands
bun install # Install dependencies
bun run typecheck # TypeScript checks
bun run lint # ESLint
bun run test # Bun tests
bun run format # PrettierDefaults
- Workspace defaults to
~/Desktop/claude-agent(change in Settings). - Anthropic API key can be provided via
ANTHROPIC_API_KEYor stored locally in Settings. - Bundled tools: bun + uv for runtime execution, portable Git/MSYS2 for Windows.
- Auto-updates: packaged builds check GitHub releases for updates; set
UPDATE_FEED_URLto override the feed if needed (e.g., staging).
Workflow
- After a series of code changes, always run: lint, typecheck, test, and format commands to ensure the code is working as expected.
- When drafting commit messages, always follow Conventional Commits format.
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.

