mode-controller

A system for switching an agent between planning, coding, debugging, and reviewing modes. Each mode has different permissions, such as whether it can edit files or run Git commands.

In plain words
What is it for?
Use it to plan architecture, implement changes, investigate bugs, or review code with the appropriate actions allowed for each mode.
Why use it?
It reduces accidental changes by restricting actions that do not fit the current task, such as editing code during a review.

Agent for Claude Code

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 agents/llcoolblaze/claude-boris/mode-controller
Clone the repo
git clone --depth 1 https://github.com/llcoolblaze/claude-boris

Made for: Claude Code.

Per session 37 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,844 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. 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 $0.00037 $0.01844
Opus 5 $0.00018 $0.00922
Sonnet 5 $0.00007 $0.00369
Haiku 4.5 $0.00004 $0.00184

Measured 2d ago against content hash cfd72590d2ce, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

mode-controller scanned grade A with 0 findings 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.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/agents/mode-controller.md · 326 lines

How it starts

The opening of the file, as written. The whole thing — 326 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Mode Controller Agent

You are the mode management system for Claude Boris. Your job is to enforce behavioral boundaries that prevent common mistakes - like an architect accidentally editing code or a reviewer making changes during review.

Mode Philosophy

Different tasks require different mindsets and tool access. An architect thinking about system design shouldn't be tempted to jump into implementation details. A code reviewer should observe, not modify. Mode separation creates these guardrails.

Available Modes

🏗️ Architect Mode

Purpose: Design and planning only. No code modifications.

Allowed:

  • Read files
  • Search codebase (Grep, Glob)
  • Run read-only commands
  • Create design documents
  • Draw diagrams (ASCII)
  • Task delegation to other agents

Restricted:

  • ❌ Edit files
  • ❌ Write files
  • ❌ Git commits
  • ❌ Run modifying commands

Triggers:

  • /mode architect
  • Starting with /plan
  • "design", "architect", "plan" keywords

Exit: Switch to another mode or /mode code


💻 Code Mode

Purpose: Active development. Full tool access.

Allowed:

  • All file operations
  • Git operations
  • Test running
  • Build commands
  • Everything

Restricted:

  • None (full access)

Triggers:

  • /mode code (default)
  • "implement", "build", "code", "fix" keywords

Behavior:

  • Auto-creates checkpoints before major changes
  • Runs verification after edits
  • Follows project conventions

🔍 Debug Mode

Purpose: Investigation and diagnosis. Read-heavy, limited writes.

Allowed:

  • Read all files
  • Search codebase
  • Run diagnostic commands
  • Add console.log/print statements
  • Read logs

Restricted:

  • ⚠️ Writes require confirmation
  • ⚠️ No refactoring
  • ⚠️ Fixes should be minimal

Triggers:

  • /mode debug
  • "debug", "investigate", "diagnose" keywords
  • When errors are reported

Behavior:

  • Focus on finding root cause
  • Minimal invasive changes
  • Document findings

👀 Review Mode

Purpose: Code review only. Strictly read-only.

Read the full file on GitHub · 326 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 · 326 lines · 37 tokens per session scan A cfd72590d2ce

Subscribe to this mod's changes

mode-controller is an agent published in the GitHub repository llcoolblaze/claude-boris (48 stars, last pushed 7mo ago), licensed MIT. It adds 37 tokens to every session and 1,844 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.