pplate
87e0b9359e
Add 4 new custom modes with BigMind guidance: - rules-bigmind/: Introspective Patrick mode (BigMind development) - rules-homelab/: Tinkerer Patrick mode (TrueNAS, Docker, infra) - rules-mcp-builder/: Craftsman Patrick mode (pi_mcps MCP servers) - rules-paisy/: Professional Patrick mode (ADP Germany payroll) Add reusable skills: - skills/assessment-first/: structured assessment.md before implementation - skills/bigmind-session-ritual/: mandatory session start/end ritual - skills/gitea-push/: conventional commit + Gitea push workflow - skills/new-mcp-server/: FastMCP scaffold procedure - skills-bigmind/, skills-homelab/, skills-mcp-builder/, skills-paisy/: mode-specific skill dirs Update existing rules: - rules-architect, rules-ask, rules-code, rules-debug, rules-orchestrator: add BigMind session guidance (search before task, announce focus, hypotheses) Add plans/MODES_AND_SKILLS_PLAN.md: full architecture document
2.2 KiB
2.2 KiB
Ask Mode Behavior — Roo Code
Persona Context — Which Patrick is asking?
The answer style and knowledge focus should match the active context:
- Homelab Patrick → ground answers in real hardware facts (TrueNAS IP, Docker stack, local storage)
- ADP/Paisy Patrick → prefer Java/Maven/Oracle/EclipseLink answers; reference compliance constraints
- MCP Builder Patrick → lean on FastMCP patterns, pi_mcps conventions, and existing server examples
- BigMind Patrick → reference schema version (v7), phase (2.7), and current tool count before speculating
Before Answering Any Question
- Search Memory:
memory_search_facts("topic keywords")+memory_search_chunks("past discussion") - Check Hypotheses: If the question touches an open hypothesis, reference it — confirm or update confidence
- Announce Focus:
memory_announce_focus(session_id, "Answering question about X", files=[], ide_hint="VS Code")
Answering Patterns
- Prefer memory over inference: If BigMind has a stored fact covering the question, lead with it and cite it explicitly
- FTS5 rule: Use 2-3 focused keywords — do NOT use long descriptive queries
- Token Efficiency: Use
memory_log_token_savewhen a memory hit avoids a full file read or web search - Honesty above comfort: If the stored fact contradicts what seems intuitive, trust the fact and explain the discrepancy
After a Substantive Answer
- Store New Facts:
memory_store_fact("category", "specific atomic fact learned")— only if genuinely new knowledge was surfaced - Flag Important:
memory_flag_important(session_id, "Key finding: X", role="assistant")— if the answer revealed something worth remembering for future sessions - Update Hypotheses: If the question resolved an open hypothesis, call
memory_resolve_hypothesis(id, status, resolution)
Special Cases
- If Patrick asks "do you remember...": use
memory_search_chunks+memory_search_factsbefore saying "no" - If the question concerns homelab infra: check stored facts first (
memory_search_facts("TrueNAS Docker")) before guessing - If uncertain: form a hypothesis (
memory_add_hypothesis) rather than stating speculation as fact