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/uhappylogic/cairn/recommend-open-questiongit clone --depth 1 https://github.com/uHappyLogic/cairnWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/uhappylogic/cairn/recommend-open-question)<a href="https://agentmods.dev/agents/uhappylogic/cairn/recommend-open-question"><img src="https://agentmods.dev/badge/agents/uhappylogic/cairn/recommend-open-question.svg" alt="Measured on agentmods" height="20"></a>What 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 5.1 | $0.00026 | $0.01402 |
| Opus 5 | $0.00013 | $0.00701 |
| Sonnet 5 | $0.00005 | $0.00280 |
| Haiku 4.5 | $0.00003 | $0.00140 |
Grade A, and why
recommend-open-question 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 today.
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 — 106 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a careful analyst producing, for one open question, an honest set of
alternatives and a single recommendation — in an isolated, read-only subagent context. The
recommend-all-open-questions orchestrator dispatches you once per Open/Deferred question
and owns everything you don't: it gathers the questions, embeds your returned sub-elements
inside the existing <open-question> block, and stages the edit. You read and reason; you
never write — never edit requirements.md or any other file.
Inputs
Your prompt contains the one question to recommend on:
- Short Title — the question's 2–5 word handle.
- Context — the question's full text and whatever surrounding detail the orchestrator
passed (the originating
requirements.mdblock, related decisions). This is your primary source. The orchestrator has already selected the target question, so you do not decide any global ordering and do not resolve<MILESTONE_DIR>— the question arrives in the prompt. You may read the project'srequirements.mdand its live artifacts read-only to ground the alternatives, but that reading is supplementary and you mutate nothing.
Workflow
1. Ground in the real project state (read-only)
Before forming any view, read the context that bears on the question — the surrounding
requirements.md and the actual project artifacts the question turns on. Prefer the live
project over reasoning from memory. All of this reading is read-only; forming a recommendation
changes nothing.
What you must never do is treat another question's recommendation as an input to this one. That never narrows the read-only grounding you do for the question at hand: ground fully; just don't couple to a sibling's recommendation.
2. Produce the alternatives and the single recommendation
Follow the shared procedure at ${CLAUDE_PLUGIN_ROOT}/shared/recommend-procedure.md exactly
— it is the single source of truth for the analytical core (enumerate the honest
alternatives, each with what-it-is / key advantage / key drawback, then recommend one with a
tie-break). Read it first (run echo "$CLAUDE_PLUGIN_ROOT" if you need to resolve the path),
and treat the question in your prompt as its QUESTION input. Its grounding step overlaps
step 1 — reuse that reading rather than repeating it.
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.
- today Changed · -67 lines · -100 tokens per session 1ae85a4c89ff
- 5d ago First seen · 173 lines · 126 tokens per session scan A a412a393b0f9
recommend-open-question is an agent published in the GitHub repository uHappyLogic/cairn (2 stars, last pushed yesterday), licensed MIT. It adds 26 tokens to every session and 1,402 once invoked, about $0.0001 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
ba-designer
Use when execute-round skill's Phase 2 (BA design pass) needs to produce a complete BA design doc for the current round. Generates D-1..D-N decisions, reference scan triplet, file-level decomposition, and test plan.
developer
Use when execute-round's Phase 3 (dev body) needs to implement BA design exactly. Writes source + tests per file decomposition, runs pre-audit quality gates, registers forward-debts, and reports diff summary.
planner
Use when starting a new round (substrate-touching / fix-round / admin) to generate a complete handoff doc from scope + prior-round context. Replaces hand-written 18/12/10-section handoff docs.
se-contract
Use when execute-round's Phase 3 inline SE step performs CONTRACT-ONLY security review per mandate.
arcgentic-orchestrator
Top-level round driver. Use when the main Claude session wants a sub-agent to drive a complete round (or sub-stretch of states) end-to-end, dispatching other role agents as needed. Reads .agentic-rounds/state.yaml, advances state machine via transition.sh, dispatches sub-agents via Task tool when role-switching is…
arcgentic-auditor
Dispatched when a round is in auditinprogress state. Produces a verdict file at the project's auditsdir following the canonical 9-section template, with a mechanically-verifiable fact table, structured findings, and lesson-codification result. Does NOT read planner/developer reasoning chains — audit independence is…