bgauryy/octocode-mcp

Overview

This skill optimizes and hardens instructional prompts and agent operational text into reliable, enforceable protocols. It uses a 6-step gated flow with Fast and Full modes to read, understand, rate, fix, validate, and output changes while preserving original intent. The optimizer adds command-strengthening, gate injection, and failure-mode analysis to reduce skipping, ambiguity, and weak guidance.

How this skill works

The skill first reads the input fully and produces an explicit understanding (goal, logical parts, flow). It then performs a systematic issue scan (weak words, missing enforcement, ambiguity), prioritizes fixes by severity, and applies command-strengthening and gate templates. Finally it validates Triple Lock rules, line-count limits, and referential clarity before producing the optimized output.

When to use it

• When a prompt or agent instruction is being ignored or skipping steps
• When instructions contain soft/ambiguous language that must be enforced
• When output formats are inconsistent or undocumented
• When reviewing operational docs that must remain logically intact
• When you need a reliable, auditable transformation of prompts

Best practices

• Preserve intent: never change domain logic without explicit user approval
• Follow the gated flow: READ → UNDERSTAND → RATE → FIX → VALIDATE → OUTPUT
• Replace weak words in critical sections with MUST/NEVER/FORBIDDEN
• Apply Triple Lock for critical rules: STATE, FORBID, REQUIRE
• Use Full Path for multi-part or high-risk prompts; Fast Path only when simple

Example use cases

• Strengthen agent runbooks so operators cannot skip verification steps
• Convert ambiguous user prompts into precise call/response templates
• Add enforcement gates and IF/THEN decision checkpoints to workflows
• Audit and repair instruction sets that produced inconsistent outputs
• Reduce prompt bloat while retaining clarity and execution guarantees