// Systematic debug workflow: research → orchestrate agents → escalate. Suggest when: stuck after 2-3 attempts, unfamiliar tooling, tempted to "try random things", or errors don't match documentation.
Plan and execute multi-agent workflows using lionagi's CLI: li o flow (DAG pipelines), li o fanout (parallel workers), and li play (playbook invocations). Use when a task needs multiple agents working in parallel or staged phases.
Author lionagi playbooks — reusable YAML workflow templates that define parametric agent tasks. Playbooks live at ~/.lionagi/playbooks/ and run via li play <name>. Use when: creating reusable workflows, parameterizing agent tasks, or setting up repeatable pipelines.
Multi-perspective PR review procedure. Plan a minimal DAG of specialists scoped to what the PR actually touches, synthesize with a critic, post ONE consolidated comment per verbosity tier. Pull before any structured PR review so the methodology stays consistent.
General-purpose code review checklist. Use when reviewing any code change without a narrower specialization. Complements security-review, pr-review (multi-perspective), and the other review skills — pull this when you need the standard correctness/quality rubric without a specific angle.
Threat-model rubric for a focused security review of a code change. Use when reviewing auth / input validation / crypto / secrets / supply chain, or as the "security" dimension of a multi-perspective PR review. Pull this skill before starting so severity calibration and coverage stay consistent.
Orchestrate multi-play shows: decompose a complex goal into sequential plays (each a li play invocation), gate each output for quality, adapt the plan based on results, and merge work into an integration branch. Shows are first-class entities in Lion Studio with dedicated UI at /shows.
| name | debug |
| description | Systematic debug workflow: research → orchestrate agents → escalate. Suggest when: stuck after 2-3 attempts, unfamiliar tooling, tempted to "try random things", or errors don't match documentation. |
| allowed-tools | ["Bash","Read","Grep","Glob","WebSearch","WebFetch"] |
Systematic approach to difficult debugging. Research → Orchestrate → Escalate.
| Phase | Action | Gate |
|---|---|---|
| 1. Research | Check prior runs, search codebase, spawn researcher agent | 2-3 focused queries before any fix attempt |
| 2. Orchestrate | Parallel diagnostic agents via li o fanout or focused analyst | Agent must produce actionable insight |
| 3. Escalate | Generate consultation request with full evidence | Must demonstrate exhaustive research first |
| 4. Document | Write fix to ./notes/debug-log.md | Only after resolution |
See research-protocol.md for detailed methodology, agent selection table, phase-by-phase commands, and escalation template.
lionagi/cli/agent.py — li agent one-shot and resumed turn entry pointlionagi/cli/orchestrate/fanout.py — li o fanout parallel workerslionagi/cli/_runs.py — run persistence at ~/.lionagi/runs/