skills.rest
adam-s
adam-sCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

instruction-tuning

Iteratively run sub-agents, inspect results, and fix .claude/ instruction files.

182|19|Updated Mar 23, 2026
Search Tags:#consistency#subagents#iteration#workflows#prompt-engineering#agent-testing#instruction-tuning
One-click install
npx skills add https://github.com/adam-s/intercept --skill instruction-tuning
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: instruction-tuning
Source: https://github.com/adam-s/intercept/tree/main/.claude/skills/instruction-tuning
Command: npx skills add https://github.com/adam-s/intercept --skill instruction-tuning

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

Use sub-agents as test subjects to iteratively improve .claude/ instruction files. Run agent → inspect → fix instructions → re-run until agents follow the protocol correctly without hints.

Core Features & Use Cases

  • Iterative instruction improvement: run agent → inspect results → fix instructions → re-run until adherence.
  • Safety and consistency: ensure agents follow protocol without hints, reducing leakage of guidance.
  • Scalable governance: apply improvements across many agents and tasks in batch.

Quick Start

Launch a fresh iteration, run sub-agents on a target site, inspect results, patch instructions, and re-run until agents consistently follow the protocol.

Frequently Asked Questions about instruction-tuning

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I fix sub-agent instruction gaps in prompt-engineering workflows?▼

To fix sub-agent instruction gaps, run the agent, inspect the results, patch the .claude/ instruction files, and re-run until the pipeline consistently follows the defined protocol without requiring hints.

What is iterative instruction tuning for agent-testing pipelines?▼

Iterative instruction tuning uses sub-agents as test subjects to improve instruction files across passes, moving from gather to build stages while updating rules and references until agents adhere to the protocol.

How do I ensure sub-agent pipelines consistently follow protocol across multiple tasks?▼

You ensure protocol consistency by iteratively inspecting agent outputs and patching instructions, applying the improvements across many agents and tasks in batch to achieve scalable governance and reduce guidance leakage.

What's the best way to scale governance across many sub-agents and workflows?▼

The best way to scale governance is to iteratively run sub-agents on a target site, inspect results, patch instructions, and apply the validated improvements across many agents and tasks in batch.

Do I need to update rules and references manually when tuning sub-agent instructions?▼

Yes, when tuning sub-agent instructions, you must update rules, agents, and references accordingly across iterative passes, ensuring all changes stay generic and are committed with clear handoff documentation.

Related Skills

Explore More Software Engineering Skills→
toobulkehtoobulkeh
7

whetstone

Run a skill on a test input, self-grade, and rewrite with A/B decisions.

Community
Advanced
sixscripts-aisixscripts-ai

ghostssh-internal-comms

Coordinate governance and handoffs across agents for GhostSSH development.

Community
Advanced
pluginagentmarketplacepluginagentmarketplace
3

continuous-monitoring

Detect adversarial activity and model drift in production AI systems.

Community
Advanced
voidrotvoidrot

ruff-recursive-fix

Apply Ruff linters and fixers iteratively to Python code.

Community
Intermediate