github-copilot-superpowers subagents.instructions.md

A set of rules for an orchestrating coding agent that delegates research, file reading, searches and certain tool calls to separate helper agents.

In plain words
What is it for?
Use it when coordinating subagents or working under these repository instructions. It helps decide which agent should handle research, file access, text searches and GitHub-related work.
Why use it?
It prevents the main agent from performing restricted tasks directly and keeps those tasks assigned to the required helper. It also explains the project’s terminology and delegation boundaries.

Instructions file for GitHub Copilot

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/kasuken/github-copilot-superpowers/subagents
Clone the repo
git clone --depth 1 https://github.com/kasuken/github-copilot-superpowers

Made for: GitHub Copilot.

Per session 1,913 This file is loaded in full into every session.
When invoked 1,913 The same file — it is already loaded in full.
Security scan A 1 finding. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.01913 $0.01913
Opus 5 $0.00957 $0.00957
Sonnet 5 $0.00383 $0.00383
Haiku 4.5 $0.00191 $0.00191

Measured 2d ago against content hash 21b6504f9cca, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

github-copilot-superpowers subagents.instructions.md scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

You must NOT use the terminal for investigation/reading/searching (e.g., `cat`, `sed`, `grep`, `rg`, `find`, `ls`, `curl`, `wget`, `python -c` to inspect files).
.github/instructions/subagents.instructions.md · 194 lines

How it starts

The opening of the file, as written. The whole thing — 194 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Subagent Instructions (SSOT)

⛔ FORBIDDEN DIRECT CALLS (Check BEFORE every MCP invocation)

If you're about to call ANY of these directly, STOP and spawn a subagent instead:

❌ FORBIDDEN (Orchestrator) Tool Pattern ✅ DELEGATE TO
Context7 MCP mcp_context7_* Research subagent
Serena MCP mcp_serena_* Research subagent
GitHub MCP mcp_github_* Research subagent
File Read read_file Research subagent
Grep Search grep_search Research subagent
Semantic Search semantic_search Research subagent
File Search file_search Research subagent
List Directory list_dir Research subagent

🚨 VIOLATION CONSEQUENCE: Direct calls to forbidden MCPs pollute orchestrator context, waste tokens, and violate the SSOT contract. If detected, abort the operation immediately.

Violation examples:

  • ❌ "I'll just quickly check Context7 for this library docs" → FORBIDDEN
  • ❌ "Let me read this small file first" → FORBIDDEN
  • ✅ "I'll spawn a research subagent to check Context7" → CORRECT
  • ✅ "I'll spawn a research subagent to read and analyze these files" → CORRECT

Before ANY tool call, ask: "Is this investigation I/O?" → If YES → SUBAGENT

Terminology

  • Investigation I/O: Any read/search/fetch operation - local files, web, APIs, MCPs that retrieve data
  • Verification I/O: Build, test, lint commands that validate changes (terminal only)
  • Context Package: Structured return from research subagent (summary + citations + next actions)
  • SSOT: Single Source of Truth - this file for delegation rules

Agent Role: ORCHESTRATOR ONLY

You are the orchestrating agent. You NEVER read files or perform investigation I/O yourself. ALL such work is done via subagents.

This file is the SINGLE SOURCE OF TRUTH (SSOT) for:

  • Absolute delegation rules and restrictions
  • Context Package contract (return format from research subagents)
  • Subagent workflow and templates
  • runSubagent tool usage

Read the full file on GitHub · 194 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 194 lines · 1,913 tokens per session scan A 21b6504f9cca

Subscribe to this mod's changes

github-copilot-superpowers subagents.instructions.md is an instructions file published in the GitHub repository kasuken/github-copilot-superpowers (16 stars, last pushed 6mo ago), licensed MIT. It adds 1,913 tokens to every session, about $0.0096 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.