Agent Orchestration
Teaches Claude when to spawn implementation agents instead of burning context in the main conversation. The core insight is simple: for multi-file changes or plan phases, spin up an agent that reads files and implements in its own context, then reports back a summary. Saves you from hitting token limits on complex features. Includes a decision table (agents for new features, skip for one-line fixes) and example prompts showing how to delegate work. Best used when you're working through a multi-phase plan or building something that touches several files. The alternative is watching your context window fill up with file contents you read once and never reference again.
- skills.sh installs
- 525
- GitHub stars
- 3.9K
- last updated
- Aug 21, 2026
Install
$ npx -y skills add parcadei/continuous-claude-v3 --skill agent-orchestration --agent claude-codeInstalls into .claude/skills of the current project.
Files
Agent Orchestration Rules
When the user asks to implement something, use implementation agents to preserve main context.
The Pattern
Wrong - burns context:
Main: Read files → Understand → Make edits → Report
(2000+ tokens consumed in main context)
Right - preserves context:
Main: Spawn agent("implement X per plan")
↓
Agent: Reads files → Understands → Edits → Tests
↓
Main: Gets summary (~200 tokens)
When to Use Agents
| Task Type | Use Agent? | Reason |
|---|---|---|
| Multi-file implementation | Yes | Agent handles complexity internally |
| Following a plan phase | Yes | Agent reads plan, implements |
| New feature with tests | Yes | Agent can run tests |
| Single-line fix | No | Faster to do directly |
| Quick config change | No | Overhead not worth it |
Key Insight
Agents read their own context. Don't read files in main chat just to understand what to pass to an agent - give them the task and they figure it out.
Example Prompt
Implement Phase 4: Outcome Marking Hook from the Artifact Index plan.
Sponsors
Agent CaffeineSponsoredClose your lid, come back to finished work
SnugSponsoredThe clipboard manager for the AI eraPromote your product hereReach 125k+ visitors monthly
WaxSponsoredBlock anything with supervision, without erasing your iPhonePromote your product hereReach 125k+ visitors monthly
knobsSponsoredSee your site on any phone, any day, in any languageDetails
- Repository
- github.com/parcadei/continuous-claude-v3
- Categories
- AI & Agent Building, Automation & Workflows
- First seen
- Jun 3, 2026
More from parcadei/continuous-claude-v3
All 108- Ast Grep Find523
- Git Commits523
- Remember523
- Repo Research Analyst522
- Review521
- Planning Agent520
- Repoprompt517
- Agent Context Isolation514
- Parallel Agents514
- Create Handoff512
- Continuity Ledger511
- Debug509
- Hooks509
- Prove508
- Compound Learnings507
- Hook Developer506
- Tldr Code506
- Commit503
- Research External503
- Test503
- Background Agent Pings502
- Debug Hooks501
- Describe Pr501
- Braintrust Tracing498
Recommended
More AI & Agent Building
self-improving-agentgrail-computer/self-improving-agentMeta-skill for making the agent self-improving. Covers updating AGENTS.md, creating new skills from repeated workflows, and deciding what to systematize. Invoke520skills.sh installs
planning-agentparcadei/continuous-claude-v3Planning agent that creates implementation plans and handoffs from conversation context520skills.sh installs
ai-sdk-uijezweb/claude-skillsBuild React chat interfaces with Vercel AI SDK v6. Covers useChat/useCompletion/useObject hooks, message parts structure, tool approval workflows, and 18 UI err517skills.sh installs
google-search-agent-teneoteneoprotocolai/teneo-skillsAI agent for Google Search Agent via Teneo Protocol. Use this skill when the user needs Google Search Agent via the bundled Teneo CLI and you need the live comm516skills.sh installs