learning_ai_common_plat/docs/ecosystem/delegation/README.md

Delegation Pack

Ready-to-send task files for Saravana, Codex, Windsurf, and Cursor. Use these files when assigning bounded ecosystem work to another coding agent.

---

Current Delegation Set

• saravana/TASK.md
• saravana/STATUS.md
• saravana/FEEDBACK.md
• codex/TASK.md
• codex/STATUS.md
• codex/FEEDBACK.md
• windsurf/TASK.md
• windsurf/STATUS.md
• windsurf/FEEDBACK.md
• cursor/TASK.md
• cursor/STATUS.md
• cursor/FEEDBACK.md
• TWO_WAY_COMMUNICATION_REFACTOR_PLAN.md
  • historical refactor blueprint for how the 3-file communication loop was introduced

---

Standard Launcher Pattern

Give an agent a one-liner in this form:

git pull origin main && read <participant>/TASK.md and <participant>/FEEDBACK.md, execute within ownership boundaries, and write progress back to <participant>/STATUS.md

Example:

git pull origin main && read /Users/saravana/BytelystAI/learning_ai/learning_ai_common_plat/docs/ecosystem/delegation/windsurf/TASK.md and /Users/saravana/BytelystAI/learning_ai/learning_ai_common_plat/docs/ecosystem/delegation/windsurf/FEEDBACK.md, execute within ownership boundaries, and update /Users/saravana/BytelystAI/learning_ai/learning_ai_common_plat/docs/ecosystem/delegation/windsurf/STATUS.md

---

Rules

• TASK.md must stay bounded
• TASK.md must name owned files or modules
• TASK.md must name do-not-touch boundaries
• TASK.md must include verification expectations
• assigned agents must update STATUS.md instead of reporting completion only in chat
• Codex reviews delegated outputs before ecosystem completion is claimed
• each participant folder must contain TASK.md, STATUS.md, and FEEDBACK.md