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/mblua/agentscommander/creating-agentsgit clone --depth 1 https://github.com/mblua/AgentsCommanderWhat 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 | $0.00000 | $0.01271 |
| Opus 5 | $0.00000 | $0.00635 |
| Sonnet 5 | $0.00000 | $0.00254 |
| Haiku 4.5 | $0.00000 | $0.00127 |
Grade A, and why
creating-agents 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 — 120 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Creating agents
For developers ready to build their own agents instead of using only the templates. Three creation paths, four required fields, one rule to follow.
What an agent is
An agent is a directory with a role-prompt file at its root. The directory IS the agent's identity. AgentsCommander selects the managed role filename for the coding agent it launches:
| Coding agent | Role file |
|---|---|
| Claude Code | CLAUDE.md |
| Codex | AGENTS.md |
| Antigravity | AGENTS.md |
| Pi | AGENTS.md |
When the agent dir lives inside an AC project at .ac/_agent_<name>/, AC promotes it to an agent matrix with optional memory/, plans/, skills/, and a canonical Role.md.
The one rule
One agent per directory tree. No nested role files. Coding agents read freely inside their working directory; a second role prompt would leak into the first agent's context, drift its behavior, and waste tokens.
If you need two agents in the same project, give each one its own sibling directory.
Path 1 — through the UI (recommended)
- Open the New Agent modal from the sidebar (the New Agent button in the project header).
- Type a name (no spaces, slashes, or NUL; lowercase kebab-case recommended).
- Pick a role template — see the role-template picker.
- Optionally enable skills if the template ships them.
- Click Create.
AC creates <project>/.ac/_agent_<name>/ with the matrix layout, writes the chosen role into Role.md, copies skills if requested, and registers the agent. The coding-agent role file (CLAUDE.md or AGENTS.md) is materialized from Role.md plus AC context at launch.
Path 2 — through the CLI
agentscommander create-agent --project MyProject --name "dev-rust" --description "Implements the Rust backend."
--project is a registered AC project folder name (paths are not accepted), --name is sanitized into a lowercase _agent_<id> folder id, and --description is written into the Role.md frontmatter and body. The command creates <project>/.ac/_agent_<id>/ with the matrix layout (memory/, plans/, skills/, inbox/, outbox/) and writes Role.md.
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 · 120 lines · 0 tokens per session scan A ba0dfe112e23
creating-agents is an agent published in the GitHub repository mblua/AgentsCommander (10 stars, last pushed 2d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,271 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.
Other agents, from other repositories
engine-implementation-executor
Execute an already-reviewed phase.rs implementation plan surgically. Receives the approved plan + scope, edits files, runs Tilt-first verification, and returns a diff summary with any judgement-call notes. Does NOT plan, does NOT review, does NOT commit. Spawned by the /engine-implementer skill.
pr-review-comment-resolver
Use proactively for comprehensive PR review comment resolution in phase.rs. Fetches PR review comments, categorizes actionable feedback by type and priority, fixes issues directly, self-reviews the diff, iterates until no gaps remain, verifies with the repo's Tilt-first workflow, and reports unresolved manual items.
mtg-rules-auditor
Audits MTG Comprehensive Rules coverage in engine code. Accepts targeted file lists or CR sections for lightweight runs, or does a full codebase sweep. Produces structured reports mapping game logic to CR rule numbers.
parser-gap-finder
Analyzes parser coverage gaps, classifies them by failure reason, and proposes prioritized parser fixes to unlock the most cards with the least code changes. Run with cargo parser-gaps data available.
git-workflow
Hermes CN 的需求与 bug 修复通常同时横跨 Desktop 与 Core 两个仓库。正式动手写代码前，两个仓库都必须先过这道预检，不要直接在 main 上改：.
PROJECT_MAP
Agent "PROJECT_MAP" from zyqzyq/Unfour, covering projectmap.md, top-level directory structure, frontend packages, @unfour/ui (packages/ui) and @unfour/command-client (packages/command-client).