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 commands/ushibo/brigade/rungit clone --depth 1 https://github.com/ushibo/brigadeWhat 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 | $0.00020 | $0.04687 |
| Opus 5 | $0.00010 | $0.02344 |
| Sonnet 5 | $0.00004 | $0.00937 |
| Haiku 4.5 | $0.00002 | $0.00469 |
Grade B, and why
run scanned grade B with 1 finding 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.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
> Workers will run in **foreground** mode — you will see their progress live and may be asked to approve some tool calls (Write, Bash for git/test runners). Approve with "Allow for session" on the first call of each kind How it starts
The opening of the file, as written. The whole thing — 412 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are executing the current sprint directly in the top-level conversation. Do not delegate orchestration to manager-agent — Claude Code subagents cannot spawn further subagents, so the top-level session must drive the wave-by-wave execution itself.
Preflight
-
Verify
.planning/tasks/ROLES.mdexists. If missing, tell the user to run/brigade:initfirst and stop. -
Verify the working tree is clean (
git status). If dirty, ask the user to commit or stash and stop. -
Confirm you are on
mainormaster. If not, ask the user to check out the base branch and stop. -
Read
.planning/config.jsonif it exists. Parseconfig.review.mode(defaultfull) andconfig.review.agents(formode: custom). Used below to decide which reviewers to spawn at the merge gate. -
Tell the user about the permission flow before starting. Print a short notice:
Workers will run in foreground mode — you will see their progress live and may be asked to approve some tool calls (Write, Bash for git/test runners). Approve with "Allow for session" on the first call of each kind; Claude Code remembers it and subsequent calls will not prompt. The main conversation is blocked while a wave runs — you can move workers to background with
ctrl+bif you need the chat free, but then preapprove the required tools in~/.claude/settings.jsonfirst.Then proceed.
Step 1 — Find the current sprint and wave
- List
.planning/tasks/sprint-*/and pick the highest-numbered sprint as current. Read its_README.md. - Read
.planning/tasks/ROLES.mdinto memory — you will need the role → agent mapping later. - Sync statuses with git history. Run
git log --onelineon the current branch. For every task spec in the sprint whosestatus:is stilltodo, check whether any commit message contains its task id (e.g.s1.3). If yes, update that task's frontmatter tostatus: doneand mark it done in_README.md. Commit the status updates to the base branch with messagechore: sync sprint statuses. - Identify the current wave: the lowest-numbered wave that still has tasks with
status: todo. If every task isdone, skip to Step 4 "Final verification".
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 · 412 lines · 20 tokens per session scan B 77bf180f4a03
run is a command published in the GitHub repository ushibo/brigade (1 stars, last pushed 1mo ago), licensed MIT. It adds 20 tokens to every session and 4,687 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
OPSX: Propose
Propose a new change - create it and generate all artifacts in one step.
create
Scaffold a new plugin with directory structure, manifests, and marketplace registration.
compose-stack
Compose a sequenced Intelligence Stack Plan for a sovereign person. Analyzes their Genius Profile + Freedom Path + stated priorities + life stage, then sequences which of the 9 layers activate first, second, third. Produces a 90-day sprint plan for building their full intelligence stack. For humans with Genius Profile…
bugfix
TDD-driven bugfix workflow: tester writes failing test (RED) → developer fixes (GREEN) → developer refactors (REFACTOR) → reviewer validates. Accepts issue number, description, or both. Auto-creates PR unless --no-pr flag is passed.
develop
Implement skill development issues with TDD-governed workflow.
dev-task
Create an agent team to develop: Developer teammate + Code-tester teammate + QA-tester teammate + Reviewer teammate + Researcher subagent → implements .dev/cdt/plans/plan-$TIMESTAMP.md in waves.