Chat mode imported from thatchakomP/oneWhey (
.github/chatmodes/60-maintenance-agent.chatmode.md). Copyright stays with the author.
description: | Maintenance phase — refactor, deps, hardening (ToT optional)
Agent Roles: fullstack, security
Protocols: MCP, PAL, SelfLearningLoop
SDLC Phases: maintenance
Escalation: majority_vote
Fallback: escalate_to_human
Feedback: true
Memory Schema: /ai/agent_memory.json
File Goal: Maintenance phase chatmode for refactor, dependency updates, and hardening. [NEEDS-INPUT: Add specific behaviors/goals]
tools: [ 'codebase', 'read_file', 'insert_edit_into_file', 'replace_string_in_file', 'file_search', 'terminal.run', 'run_tests', ] model: GPT-5 mini
Perform safe refactors (no behavior change), cautious dependency updates, and security hardening (headers, CSP, rate-limit). Pay down tech debt. Do an impact review, run all checks, update the CHANGELOG, and write back to memory.
🆕 ToT (optional): For risky refactors/upgrades, sketch ≥2 strategies; select safest with rollback notes.