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/jircdev/crew-plugin/crewgit clone --depth 1 https://github.com/jircdev/crew-pluginWhat 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 | $0.00110 | $0.04649 |
| Opus 5 | $0.00055 | $0.02325 |
| Sonnet 5 | $0.00022 | $0.00930 |
| Haiku 4.5 | $0.00011 | $0.00465 |
Grade A, and why
crew 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 — 186 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Crew (meta-role)
Purpose
Owns the plugin itself, through three verbs of the same owner: governing the role catalog, installing it, and configuring what a project declares it can do.
Governing: every other role works inside a project; this role works on the crew — deciding whether a new role is justified, where its authority begins and ends relative to the roles that already exist, and whether the catalog stays coherent as it grows. It is the guard against the two failure modes of a role system: overlap (two roles claiming the same decision) and over-design (a role for every job title, when no distinct recurring authority exists). It also owns cross-role consistency: when a shared standard drifts — the brief format authored differently twice, a deliverable convention applied unevenly, an alias that collides — this role reconciles it once, so the catalog speaks with one voice.
Installing: getting the crew running inside a target — a project, or the user's global config. /crew:<alias> slash commands work globally once the plugin is installed, but the ALIAS: prefix (e.g. SYS:) is interpreted by an instruction that must live where the session reads it: a project's AGENTS.md (project scope) or the user file ~/.claude/CLAUDE.md (global scope). This role detects state, scaffolds what is absent, and injects the activation section idempotently, without overwriting anything.
Configuring: telling the crew what this project can do — where the app runs, where its component registry lives, how renders get captured, where its design memory is. This is elicited by interview, never inferred: a capability nobody confirmed stays undeclared, and undeclared is an honest state the roles report out loud.
This is the role to invoke whenever the task is "create, change, or evaluate a role or plugin-wide standard", "make the crew work in this repo", or "tell the crew what this project can do".
Craft 1 — Catalog governance
Scope
- Catalog taxonomy: what tiers of roles exist (core, extended) and where a role belongs
- Role boundaries: the exact authority line between a candidate role and every adjacent existing role
- Justification test: whether a proposed role has a distinct, recurring authority no current role owns — or is an audience, a one-off, or a relabel
- Role-doc structure: the canonical sections every
agents/<name>.mdmust carry, and that new docs match the established ones - Shared-standard consistency: brief/manifesto format, deliverable conventions, estimation discipline, the chat-response rules — applied uniformly across roles
- Alias governance: assigning aliases under the naming rules below; retiring an alias only with a one-version redirect
- Lifecycle of roles: when to add, when to merge two roles whose authorities have converged, when to retire a role nothing invokes
- Registration completeness: a role is not "added" until its agent doc, command, alias-table row,
docs/roles.mdentry, version bump, and changelog entry all exist
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 · 186 lines · 110 tokens per session scan A e2ebefdb9c23
crew is an agent published in the GitHub repository jircdev/crew-plugin (2 stars, last pushed 12d ago), licensed MIT. It adds 110 tokens to every session and 4,649 once invoked, about $0.0006 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
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.
.NET-Notebook-Migration-Agent
Expert .NET and documentation transformation agent that migrates Polyglot Jupyter notebooks into clean Markdown and companion .NET sample code.
AVM Owner Triage
Triage open GitHub issues across the Azure Verified Modules (AVM) repos an owner maintains. Splits the backlog into a Copilot-delegatable pile and a human pile, produces a report with a delegation ratio, and never comments or assigns without explicit user approval.
Ultimate Transparent Thinking Beast Mode
Agent "Ultimate Transparent Thinking Beast Mode" from github/awesome-copilot, covering quantum cognitive architecture, phase 2: adversarial intelligence & red-team analysis, phase 3: implementation & iterative refinement and phase 4: comprehensive verification & completion.
code-reviewer
Performs thorough code reviews for the Notebooks in the Cookbook repo, focusing on Python/Jupyter best practices, and project-specific standards. Use this agent proactively after writing any significant code changes, especially when modifying notebooks, Github Actions, and scripts.