Compare session list before and after fork to reliably detect which
session is the newly created one. Avoids relying on parsing output
that may not contain session ID.
Create placeholder session file and add to index BEFORE running
opencode. This ensures we have a record even if opencode takes
long time or times out. Update with real session ID after fork.
The --fork flag outputs the new session ID. Parse that instead of
relying on session list which may return wrong session when multiple
exist. Added fallback to session list parsing.
- Add --debug flag to start/resume for verbose opencode output
- Use stdbuf -oL to unbuffer stdout for real-time display
- Capture debug logs to ~/.kugetsu/sessions/<id>/debug.log
- Add --debug to stop/destroy for viewing logs before actions
- Position-agnostic flag parsing (--debug can appear anywhere in args)
- skills/kugetsu/SKILL.md: Agent skill documentation following agentskills.io spec
- skills/kugetsu/scripts/kugetsu: Shell wrapper for opencode session management
- Commands: start, list [--all], resume, stop, help
- State tracking: used → idle (graceful) or left (interrupted)
- Auto-fill message on resume
- Confirmation prompt when resuming used session
- skills/kugetsu/scripts/kugetsu-install.sh: Installation script for users
Implements Phase 1 of issue #11 - basic session management layer
for remote agent control without Hermes dependency.
- Installation via curl script (with --skip-setup for CI)
- API key configuration for 9+ LLM providers
- OpenCode delegation via terminal() wrapper pattern
- Git worktree isolation per-issue workflow
- Reference to existing opencode-worktree skill
Related to issue #1
Document:
- How to detect contamination via git log and branch --contains
- Prevention with explicit base: git checkout -b new-branch main
- Fix using git rebase --onto
- Force push with --force-with-lease for safety
Addresses Issue #1 comment 281
- Adds skills/opencode-worktree/ with SKILL.md and opencode-worktree.sh
- Creates unique git worktree per session (e.g. session-20260327-a1b2c3-refactor-auth)
- Cleans up stale worktrees on every launch
- Branch always based on main
- User can source directly or copy to PATH
This tool tests the practical limits of parallel agent execution
by spawning N concurrent opencode run tasks and measuring:
- Response time
- CPU and memory usage
- Success/failure rates
Includes both bash (run_test.sh) and Python (parallel_capacity_test.py)
implementations with full metrics collection and reporting.
Fixes#3
