---

name: fresh-context-orchestrator description: Fight context rot by spawning fresh sub-agents per task with quality gates for scope reduction and security. Use when working on fresh-context sub-agent orchestrator.

Fresh-Context Sub-Agent Orchestrator

Wraps the GSD (get-shit-done) sub-agent pattern so long Claude Code sessions do not degrade around the 30-minute mark. Each spawned agent gets a clean window scoped to only what it needs. A supervising planner catches silently dropped requirements. Optional autonomous mode lets you hand in a spec and walk away. Use when a session is long enough that context bloat starts costing you output quality.

What you get

• Task-splitter prompt
• Sub-agent spawn recipes
• Scope-reduction detector
• Security threat-model check
• Per-agent activity log
• Commit-on-success option

Customize your output

• Autonomy level: manual approval, semi-autonomous, hands-off
• Token budget cap: 50k, 200k, 1M
• Task granularity: coarse, medium, fine
• Commit on success: yes, no
• Recovery on failure: retry, escalate, halt
• Concurrency: 1, 3, 5

Example output

'A five-hour build broken into 14 sub-agent runs, each with a clean context, ending in a green PR.'

Best for

Solo devs, agency builders, indie SaaS operators.

---

Note: Costs tokens. Quality and hours play, not a savings play. High confidence.

Source: AI Agencies Creator (unnamed) · https://www.youtube.com/watch?v=eRS3CmvrOvA Category: AI Systems · Priority: P4