habv4-token-efficient

Activation

This skill activates automatically for all tasks in this repository.

Instructions - Strict Workflow

1. Plan First (always):

   • Create a brief 3-6 bullet plan.
   • Decide which files/directories are truly needed.
   • Summarize relevant files before full read.

2. Progressive Context Loading:

   • Use ls, tree, grep -r or partial reads first.
   • Only request full file content if the summary confirms it's required.
   • Maintain and update a Persistent Session Summary (intent, decisions, changes, next steps).

3. Model & Reasoning:

   • Default: Fast model + low/medium reasoning for analysis and small edits.
   • Switch to high-reasoning only when generating patches or modifying boot logic.

4. Output Format (Token-Saving):

   • Show only changed code + minimal explanation.
   • Use concise diffs when possible.
   • End every response with: Current Session Summary: [2-4 bullets]

Success Criteria

• Task completed with ≤ 40-60% of baseline token usage.
• All changes verified via diagnose_iso_repodata.sh or equivalent.
• Persistent summary updated.

Never

• Load entire directories or large files without summarization.
• Give long explanations unless explicitly asked.