bala

name	bala
description	Apply the Buddhist Five Powers as a secular engineering balance check for confidence, effort, awareness, focus, and causal judgment.

Bala 5 Engineering Check

Use the Buddhist Five Powers (พละ 5) as a practical operating check for AI-assisted engineering work. This skill is secular by default: it translates the five powers into observable subagent behavior, not religious instruction.

Workflow

Run this lens directly in the current conversation; do not route it to a planning, implementation, or verification worker.

1. Inspect only the supplied scope and the minimum relevant repository context.
2. Score each power from observable evidence, not confidence or rhetoric.
3. Identify the single imbalance most likely to create waste or harm.
4. Recommend one small, reversible next action.
5. Name the test, log, review signal, or user outcome that would prove it worked.

Return a concise balance report. Do not change files unless the user separately asks for implementation.

Five Powers Mapping

1. Saddhā / Confidence (ศรัทธา) - confidence is calibrated by evidence.

   • Good sign: goal, user value, and acceptance criteria are clear.
   • Bad sign: blind optimism, vague success, or acting because the agent sounds sure.
   • Ask: "What evidence makes this worth doing now?"

2. Viriya / Energy (วิริยะ) - effort is steady and pointed at the next smallest useful action.

   • Good sign: the next action is small, reversible, and verifiable.
   • Bad sign: thrashing, huge rewrites, or opening too many fronts.
   • Ask: "What is the smallest useful move that reduces uncertainty?"

3. Sati / Mindfulness (สติ) - attention stays aware of context, constraints, and current state.

   • Good sign: dirty files, assumptions, risks, and user constraints are visible.
   • Bad sign: forgetting prior decisions, overwriting human work, or ignoring safety boundaries.
   • Ask: "What must stay in awareness before touching anything?"

4. Samādhi / Concentration (สมาธิ) - focus is narrow enough to finish.

   • Good sign: one objective, one branch of work, minimal interruptions.
   • Bad sign: parallel tasks that touch the same files, context sprawl, or premature polish.
   • Ask: "What should we stop doing until this is verified?"

5. Paññā / Wisdom (ปัญญา) - decisions are based on causal understanding and tradeoffs.

   • Good sign: root cause, alternatives, risks, and rollback are explicit.
   • Bad sign: patching symptoms, copying patterns blindly, or hiding uncertainty.
   • Ask: "What do we know, what do we not know, and what would change the decision?"

Output Format

## Bala 5 Check
- Saddhā / Confidence: <green|yellow|red> - <evidence>
- Viriya / Energy: <green|yellow|red> - <evidence>
- Sati / Mindfulness: <green|yellow|red> - <evidence>
- Samādhi / Concentration: <green|yellow|red> - <evidence>
- Paññā / Wisdom: <green|yellow|red> - <evidence>

Most likely imbalance: <one sentence>
Smallest next action: <one concrete action>
Proof it worked: <test/log/review signal>

Common Uses

• Before planning: check whether the ask is grounded enough to plan.
• Before implementation: prevent overbuilding and pick the smallest implementation slice.
• Before review: separate confidence from evidence and focus the review lens.
• During debugging: slow down guessing and return to root-cause discipline.
• During multi-subagent work: decide whether parallelism is safe or whether focus is needed.

Evidence Receipt

Return spk.evidence/v1 with the five ratings and supporting evidence, the primary imbalance, smallest next action, proof signal, risks, and next action.

Guardrails

• Do not use Bala 5 as a motivational speech. Translate every point into observable engineering behavior.
• Do not shame the user or agent. Report imbalance as a workflow signal.
• Do not add extra process if the next action is already obvious and safe.
• Prefer one concrete next action over a long checklist.