README

Documentation for a group of specialized sub-agents used by a Claude Code workflow. A sub-agent is a separate agent session assigned to a focused job, such as writing tests or reviewing code.

In plain words
What is it for?
Use it as a reference for agents that write tests, review files, fix reported bugs, or audit a whole codebase for inconsistencies and broken links.
Why use it?
It explains which agents exist, what starts them, and how they differ from slash commands. Separate contexts can support parallel work without filling the main session with every detail.

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/caspian-sun/claude-code-workflow/readme
Clone the repo
git clone --depth 1 https://github.com/Caspian-Sun/claude-code-workflow

Made for: Claude Code.

Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 901 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 $0.00000 $0.00901
Opus 5 $0.00000 $0.00451
Sonnet 5 $0.00000 $0.00180
Haiku 4.5 $0.00000 $0.00090

Measured 2d ago against content hash 2191b5302607, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

README 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/README.md · 73 lines

How it starts

The opening of the file, as written. The whole thing — 73 lines — stays where its author put it; the contents beside it link to each section on GitHub.

agents/ — Sub-Agent Definitions

Specialized agents spawned by main commands via the Agent tool, each running in an independent context, supporting parallelism without polluting the main session.

File Manifest

Agent Responsibility Spawned By
test-writer.md Generate tests for specified source files (reads @rules → writes it() per rule) After /code completes / /test for multi-module parallelism
code-reviewer.md Read-only review of a file/directory, outputs issue list /review splitting large directories / pre-PR self-check
bug-fixer.md Fix a single bug (from bug-report) /fix for parallel handling of multiple bug reports
meta-auditor.md God-mode meta-audit, scans the entire codebase for inconsistencies/drift/dead links, outputs report to docs/retrospectives/ Triggered by /meta-audit command

Sub-Agents vs. Slash Commands

Slash Commands (commands/) Sub-Agents (agents/)
Runs in Main context Independent context (created at spawn time)
Parallel No, serial execution Yes, multiple agents can run concurrently
Memory Full main session visible Only sees the passed-in prompt
Output Written directly to main session Returns a summary to main agent
Best for User-triggered workflows Independent sub-tasks spawned inside main commands

Why sub-agents:

  1. Parallel acceleration — 5 independent bugs reported in tests → spawn 5 bug-fixer agents in parallel, 5× faster than serial
  2. Protect main context — reviewing a large directory reads many files; stuffing that into main context quickly overflows; let sub-agents run and return summaries
  3. Responsibility isolation — sub-agents only see the passed-in prompt, free from main session pollution, producing more focused output

Agent MD File Format

---
name: <agent-name>                    # Unique identifier, matches filename
description: <one sentence> — when to spawn    # Main agent uses this to decide whether to use it
tools: [Read, Edit, Write, Bash, Grep, Glob]  # Optional, inherits by default
---

# <agent-name>

<Complete system prompt: role / input / execution steps / output format / boundaries>

Read the full file on GitHub · 73 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 · 73 lines · 0 tokens per session scan A 2191b5302607

Subscribe to this mod's changes

README is an agent published in the GitHub repository Caspian-Sun/claude-code-workflow (10 stars, last pushed 5d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 901 tokens. 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 2026-08-31.