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/thewaywithin/bos-ai/coordinatorgit clone --depth 1 https://github.com/TheWayWithin/BOS-AIWhat 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 | $0.00055 | $0.35127 |
| Opus 5 | $0.00028 | $0.17563 |
| Sonnet 5 | $0.00011 | $0.07025 |
| Haiku 4.5 | $0.00006 | $0.03513 |
Grade C, and why
coordinator scanned grade C with 3 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.
Asks the agent to reveal its instructionsmediumSystem prompt leakage
Directions to print, repeat or translate the system prompt extract configuration the operator did not intend to expose.
5. IMMEDIATELY use Task tool with subagent_type='strategist' INCLUDING context preservation AND structured output instructions - WAIT for response Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
**The counters are on disk, not in your head (T-363).** `.claude/scripts/mission-state.py` owns Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
**Fabricating test output is a critical failure** — worse than skipping the test. *"Tests not run because the dev server needs manual start"* is honest and useful. *"Live curl on :3457/health → ok"* written without havin The source is not reproduced here
No licence file
A repository with no LICENSE is all rights reserved by default, so the body is not copied here. The metadata, the measurements and the link are.
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.
- 2d ago First seen · 3,092 lines · 55 tokens per session scan C 2bc152e094b6
coordinator is an agent published in the GitHub repository TheWayWithin/BOS-AI (7 stars, last pushed 10d ago), with no licence file. It adds 55 tokens to every session and 35,127 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it C with 3 findings (asks the agent to reveal its instructions, reads agent configuration directories, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other agents, from other repositories
orchestrator
A specialized agent that implements the Orchestrator-Workers pattern. It analyzes complex requests, breaks them down into atomic tasks, delegates them to the most appropriate specialist agents or tools, and synthesizes the results. Use this agent for "vague" or multi-step requests. For durable team design and capacity…
context-manager
Manages context across multiple agents and long-running tasks. Use when coordinating complex multi-agent workflows or when context needs to be preserved across multiple sessions. Must be used for projects exceeding 10k tokens.
help
Use this screen to check which agents are alive, what they are running, and where work is currently landing.
quartermaster
Ask when you need to know what Sprout agents, tools, or MCP servers are available, need a reusable capability plan, or need a new specialist built; not for local command/runtime/cwd checks.
editor
Ask to create named files or make targeted edits — acts directly when targets are decisive, reads only when context is missing.
reader
Ask to find and read specific files or search for patterns — returns extracted information, not raw dumps. Use for quick targeted lookups, not broad exploration.