skills/jules/SKILL.md
Jules is SWAL's autonomous coding agent from Google. Use via Jules CLI or GitHub issues with `jules` label. Jules works ONLY in iberi22/* repos. Updated 2026-04-15.
npx skillsauth add iberi22/swal-skills julesInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
⚡ SWAL Agent System — Updated 2026-04-15
# List ALL Jules sessions (PRIMARY command)
jules remote list --session
# List connected repos
jules remote list --repo
# Create new session
jules new --repo iberi22/<REPO> "task description"
jules new --repo iberi22/<REPO> --parallel 3 "task"
# Pull session result (view what Jules did)
jules remote pull --session <ID>
# Pull AND apply patch to local repo
jules remote pull --session <ID> --apply
# Teleport (clone repo + checkout + apply patch)
jules teleport <SESSION_ID>
jules remote list --repo
jules login
Opens browser for Google OAuth.
jules logout
# List all sessions (MOST IMPORTANT)
jules remote list --session
# Show session details
jules remote pull --session <ID>
# Apply session patch to local repo
jules remote pull --session <ID> --apply
# Clone repo and apply session changes
jules teleport <SESSION_ID>
# Create new session
jules new --repo iberi22/<REPO> "task description"
| Status | Meaning | Action |
|--------|---------|--------|
| In Progress | Jules está trabajando | Esperar |
| Awaiting User F | Jules pidió input y espera | Revisar, extraer parcial, eliminar y relanzar |
| Completed | Listo para review | jules remote pull --apply |
| Failed | Error | Revisar, eliminar y recrear issue |
# Direct task assignment
jules new --repo iberi22/<REPO> "implement feature X"
# Multiple parallel sessions
jules new --repo iberi22/<REPO> --parallel 3 "write unit tests"
ONLY the jules label triggers Jules!
# 1. Create issue
gh issue create --repo iberi22/<REPO> --title "[feat] Description" --body "Problem, solution, files"
# 2. Add jules label
gh issue edit <NUM> --repo iberi22/<REPO> --add-label jules
# 3. Jules picks it up automatically
@jules-ai mention = NOT a triggerjules label = ONLY triggerPROTOCOL:
Revisar qué hizo Jules antes de detenerse:
jules remote pull --session <ID>
Extraer trabajo parcial — código generado está en el output
Eliminar sesión — desde https://app.jules.ai o dejar que expire
Analizar por qué se detuvo:
Relanzar con mejor contexto:
jules new --repo iberi22/<REPO> "task + contexto adicional"
jules remote list --session
| Filter | Command |
|--------|---------|
| In Progress | jules remote list --session (look for "In Progress") |
| Awaiting User F | jules remote list --session (look for "Awaiting User F") |
| Completed | jules remote list --session (look for "Completed") |
| Failed | jules remote list --session (look for "Failed") |
# Apply to local repo
jules remote pull --session <ID> --apply
# Or teleport (clone fresh)
jules teleport <SESSION_ID>
| Project | Repo | Status | |---------|------|--------| | OrionHealth | iberi22/OrionHealth | ✅ Active | | Cortex | iberi22/cortex-1 | ✅ Active | | Synapse Protocol | iberi22/synapse-protocol | ✅ Active | | ManteniApp | iberi22/manteniapp | ✅ Active | | WorldExams | iberi22/worldexams | ✅ Active | | GOS | iberi22/gastronomic-os | ✅ Active | | Gestalt-Rust | iberi22/gestalt-rust | ✅ Active |
⚠️ Jules ONLY works in
iberi22/*repos.
# 1. ASSIGN
jules new --repo iberi22/OrionHealth "implement feature X"
# 2. MONITOR (check every few minutes)
jules remote list --session
# 3. When Completed, APPLY
jules remote pull --session <ID> --apply
# 4. Push to origin
git push origin develop
Jules handles complex async coding tasks (large refactors, new features, PR reviews). Gestalt Swarm handles fast parallel CLI operations (scans, analysis, bulk operations).
User: "Jules, analiza gestalt-rust para bugs y correcciones"
↓
Jules (async) → Planning + code review
Gestalt Swarm (sync) → Parallel exec: rg, cargo, git, curl
↓
Jules → Implement fixes based on Swarm results
↓
PR + Merge
| Task | Use |
|------|-----|
| Large refactors, new features | Jules (jules new --repo iberi22/...) |
| Bug finding, security audit | Gestalt Swarm (swarm_bridge.py --goal "...") |
| Comprehensive analysis | Jules + Gestalt Swarm combined |
| Quick status checks | Gestalt Swarm only |
| Deep coding, PR review | Jules only |
# Full analysis (smart agent selection)
python E:\scripts-python\gestalt-rust\swarm_bridge.py --goal "comprehensive security audit" --json
# Quick status
python E:\scripts-python\gestalt-rust\swarm_bridge.py --goal "git status" --agents "git_analyzer,git_status" --json --quiet
# Dry run (preview agents)
python E:\scripts-python\gestalt-rust\swarm_bridge.py --goal "analyze codebase" --dry-run
# Custom agents
python E:\scripts-python\gestalt-rust\swarm_bridge.py --goal "check dependencies" --agents "dep_check,cargo_check" --json
# Streaming mode
python E:\scripts-python\gestalt-rust\swarm_bridge.py --goal "scan files" --watch --timeout 30
# 1. Run Gestalt Swarm for fast analysis
python swarm_bridge.py --goal "security audit" --max-agents 10 --json
# 2. Based on results, assign to Jules
jules new --repo iberi22/gestalt-rust "fix security issues found: [paste from swarm output]"
# 3. Monitor Jules
jules remote list --session
# 4. When complete, apply
jules remote pull --session <ID> --apply
jules login
# Check what's happening
jules remote pull --session <ID>
# If truly stuck (Awaiting User F), let it expire or delete from app.jules.ai
jules labelRepo must be clean before sending to Jules
git status # must be "nothing to commit"
git checkout develop
Provide clear context in the task description
Break large tasks into smaller ones (1 feature = 1 issue)
Monitor sessions — don't let them sit in "Awaiting User F"
Apply completed sessions promptly before they expire
Use Gestalt Swarm for pre-flight checks before assigning to Jules
Combined workflow for maximum agentic power: Swarm first, then Jules based on results
| Agent | For |
|-------|-----|
| code_analyzer | Ripgrep patterns in .rs files |
| dep_check | Cargo tree dependencies |
| cargo_check | Cargo check compilation |
| git_analyzer | Git log history |
| git_status | Working tree status |
| file_scanner | List files in directory |
| log_parser | Find ERROR in logs |
| security_audit | Find TODO/FIXME/unsafe |
| find_todos | Find TODO/FIXME/HACK |
| api_tester | Test HTTP endpoints |
| metrics | Cargo tree stats |
| env_check | Environment variables |
| doc_gen | Find documentation files |
Last updated: 2026-04-15 Jules = Google's autonomous AI coding agent Gestalt Swarm = SWAL's parallel execution bridge
testing
Xavier2 as the central context engine for SWAL - intelligent memory, decision-making, and context orchestration. Xavier2 is the CEO brain that stores memories, coordinates agents, and maintains project state.
tools
Delegate coding tasks to Codex, Claude Code, or Pi agents via background process. Use when: (1) building/creating new features or apps, (2) reviewing PRs (spawn in temp dir), (3) refactoring large codebases, (4) iterative coding that needs file exploration. NOT for: simple one-liner fixes (just edit), reading code (use read tool), thread-bound ACP harness requests in chat (for example spawn/run Codex or Claude Code in a Discord thread; use sessions_spawn with runtime:"acp"), or any work in ~/clawd workspace (never spawn agents here). Claude Code: use --print --permission-mode bypassPermissions (no PTY). Codex/Pi/OpenCode: pty:true required.
testing
Validador automático para WorldExams. Verifica integridad técnica y calidad pedagógica, activando regeneración automática si es necesario.
tools
Generador de bundles de preguntas ICFES Colombia (Matemáticas, Lectura Crítica, Ciencias, Sociales, Inglés) para grados 6, 9 y 11 usando MiniMax MCP.