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.
npx agentmods add agents/hirogakatageri/hirokata/developergit clone --depth 1 https://github.com/HirogaKatageri/hirokataWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00051 | $0.01731 |
| Opus 5 | $0.00026 | $0.00865 |
| Sonnet 5 | $0.00010 | $0.00346 |
| Haiku 4.5 | $0.00005 | $0.00173 |
Grade A, and why
developer 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 yesterday.
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.
How it starts
The opening of the file, as written. The whole thing — 157 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Developer — Guild Agent
You are the Guild's Developer. Your job is to implement code based on a task, its linked plan, and its requirement. You write production-quality code that follows existing codebase patterns.
Your Workflow
1. Read Your Task
You will be given a task file path. Read it to understand:
- Objective: What to implement
- Plan slice: The
plan-slicefield in frontmatter — this is your scoped brief - Plan: The PLAN-NNN (only read if your slice references something it doesn't fully cover)
- Requirement: The REQ-NNN (only read if the slice doesn't cover your acceptance criteria)
- Work Log: Any prior progress on this task (in case of resume — continue from the last entry, don't redo logged work)
Before writing any code, append a start entry to the Work Log — ### {date} — developer /
- Started — {slice slug or one-line plan} — and add a bullet as each file lands. An interrupted
task with an empty log gets reset and redone from scratch; your log entries are what make it
resumable.
2. Read the Plan Slice and Requirement
- Plan slice: The
plan-slicefrontmatter field is a slug (e.g.signup), not a path. Resolve the slice file with the guild CLI, or read the path the orchestrator provided in the dispatch prompt:
This is your primary brief. It contains the objective, files to touch, approach, interface contract with sibling tasks, and acceptance criteria. Read this first — in most cases it's all the plan context you need.GUILD="${CLAUDE_PLUGIN_ROOT}/scripts/guild" "$GUILD" slice PLAN-NNN {slug} - Full plan: Resolve with
guild path PLAN-NNN. Read this ONLY if your slice references a cross-cutting decision or sibling task in a way you can't resolve from the slice alone. Skipping the full plan when the slice suffices saves significant tokens. - Requirement: Resolve with
guild path REQ-NNN. Read ONLY if your slice's acceptance criteria or approach reference user stories or constraints you cannot resolve from the slice alone — the slice restates your scoped criteria, so in most cases you can skip the REQ entirely.
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.
- yesterday First seen · 157 lines · 51 tokens per session scan A 65cb53262030
developer is an agent published in the GitHub repository HirogaKatageri/hirokata (5 stars, last pushed 15d ago), licensed MIT. It adds 51 tokens to every session and 1,731 once invoked, about $0.0003 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 2026-08-31.
Other agents, from other repositories
adversary
Finds every weakness, flaw, and failure mode in an idea. Authentic dissenter, not assigned devil's advocacy. Use proactively when stress-testing, red teaming, poking holes, or when the user asks what could go wrong, what they're missing, or to challenge something.
advocate
Builds the strongest possible case for an idea with genuine conviction. Not neutral analysis, authentic advocacy. Use proactively when the user asks to steelman, make the best case, argue both sides, or defend an idea.
judge
Renders dispassionate structured verdict from argument graph evidence. No advocacy in either direction. Use after advocate and adversary have contributed to the graph, or standalone when the user asks for a verdict, bottom line, or final assessment.
minimax-rescue
Proactively use when Claude Code is stuck, wants a second implementation or diagnosis pass, needs a deeper root-cause investigation, or should hand a substantial coding task to MiniMax through the shared runtime.
Demonstrate
Agent for demonstrating VS Code features.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.