auto
Automatically converge from goal to A-grade Seed and execute it
Cancel stuck or orphaned executions
$ npx -y skills add Q00/ouroboros --skill cancel --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/cancelContext preview
The summary Claude sees to decide when to auto-load this skill.
Cancel stuck or orphaned executions
name: cancel description: "Cancel stuck or orphaned executions" aliases: [kill, abort]
Cancel stuck or orphaned executions by session ID, cancel all running sessions, or interactively pick from active executions.
/ouroboros:cancel # Interactive: list active, pick one /ouroboros:cancel <execution_id> # Cancel specific execution /ouroboros:cancel --all # Cancel all running executions
**Trigger keywords:** "cancel execution", "kill session", "stop running", "abort execution"
This skill interacts **directly with the event store** (not via MCP tool) to emit cancellation events. It uses the CLI command under the hood.
Three modes:
1. **Bare (no args)**: Lists all active (running/paused) executions in a numbered table and prompts you to pick one to cancel 2. **Explicit (`execution_id`)**: Cancels the specified execution immediately 3. **`--all` flag**: Cancels every running or paused execution at once
When the user invokes this skill:
1. Determine which mode to use:
2. Run the appropriate CLI command using Bash:
**Bare mode** (interactive):
ouroboros cancel execution
This will list active executions and prompt for selection.
**Explicit mode** (specific execution):
ouroboros cancel execution <execution_id>
**Cancel all mode**:
ouroboros cancel execution --all
**With custom reason**:
ouroboros cancel execution <execution_id> --reason "Stuck for 2 hours"
3. Present results to the user:
4. End with a next-step suggestion:
Only sessions in `running` or `paused` status can be cancelled. Sessions that are already `completed`, `failed`, or `cancelled` are skipped with a warning.
If the configured runtime EventStore does not exist:
No Ouroboros database found at the configured runtime path. Run an execution first with: /ouroboros:ouroboros-run
User: cancel that stuck execution > ouroboros cancel execution Active Executions ┌───┬──────────────────┬──────────────┬─────────┬─────────┬──────────────┐ │ # │ Session ID │ Execution ID │ Seed ID │ Status │ Started │ ├───┼──────────────────┼──────────────┼─────────┼─────────┼──────────────┤ │ 1 │ sess-abc-123 │ exec-001 │ seed-42 │ running │ 2024-01-15 │ │ 2 │ sess-def-456 │ exec-002 │ seed-99 │ paused │ 2024-01-14 │ └───┴──────────────────┴──────────────┴─────────┴─────────┴──────────────┘ Enter number to cancel (1-2), or 'q' to quit: 1 Cancel session sess-abc-123 (running)? [y/N]: y ✓ Cancelled execution: sess-abc-123 ◆ Cancelled — use `ooo status` to verify, or `ooo run` to start fresh
Your final response MUST end with exactly one breadcrumb footer line:
◆ <current state> → next: <recommended action>
Derive `<current state>` from live session state via `ouroboros_session_status` when that MCP projection is available; otherwise derive it from this skill's actual outcome. Never use a linear `Step N of M` footer because Ouroboros is an evolutionary loop. When the next action is genuinely a choice, list 2-3 honest options in the `next:` clause. The breadcrumb line must be the last line of the response.
Agent OS: the agent gets smarter on its own. We just hold the line: Interview-gated, staged evaluation, budgeted evolution loop. MCP server, 14 runtimes: Claude Code, Codex CLI, Gemini CLI, OpenCode, Copilot, Kiro and more.
Repo: Q00/ouroboros
Automatically converge from goal to A-grade Seed and execute it
Scan and manage brownfield repository/worktree defaults for interviews
Open or drive the Ouroboros settings GUI (browser, TUI, or conversational fallback)
Evaluate execution with three-stage verification pipeline