A third-party report says Roo Code was sunset on 2026-05-15; its docs still load without a notice. EndoStack installs skills and commands for it, but it gets the least testing of any agent here.
Files, one project:
./install.sh --scope project --agents roo-codeFiles, every project:
./install.sh --scope user --agents roo-codeRun the installer from a clone of the EndoStack repository. Add --dry-run to see every file it would write, --link to symlink instead of copy, and --uninstall (with the same --scope) to remove exactly what it wrote. On Windows use install.ps1 with the same options.
| What | Path |
|---|---|
| Skills | .agents/skills/<name>/ or ~/.agents/skills/<name>/ |
| Commands | .roo/commands/<name>.md or ~/.roo/commands/<name>.md, one per command-style skill |
| Rules | A marked block in the project AGENTS.md (unverified) |
The installer records every file it writes, with a hash, in .endostack/install-manifest.json (project) or ~/.endostack/install-manifest.json (user). It never overwrites a file it did not write.
EndoStack skills use neutral verbs. In this agent they mean:
| EndoStack verb | Tool in Roo Code |
|---|---|
| Spawn a subagent | none |
| Ask the user | not documented |
| Track a todo list | not documented |
| Run a shell command | not documented |
| Read a file | not documented |
| Edit a file | not documented |
| Search the codebase | not documented |
| Fetch a URL | not documented |
| Load a skill | skill |
| Transcripts | Per task, in the editor's extension storage under rooveterinaryinc.roo-cline/tasks/<task-id>/ (unverified). |
"(unverified)" marks a name or path taken from a third-party source or prior knowledge rather than the vendor's docs. Where a verb has no tool, skills fall back as they describe: plain-text questions, a checklist in the reply, or doing subagent briefs one after another.
No subagents, so there is no per-subagent model. Every role runs in the main chat on the session model.
Role assignments live in ~/.endostack/models.toml or the project's .endostack/models.toml, written by setup-endostack.
- No subagents. Skills run every subagent brief one after another in the main chat, which uses more context and takes longer.
- Panels (
arena,interrogate,security-review) lose independence: one model writes every seat's answer in one context, so treat their agreement as weaker evidence.
Non-interactive mode (used by make-bot-ui for a local bot): not documented.