Borrowing it
Nothing to install: this file belongs to smithersai/smithers. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/smithersai/smithers/main/.agents/skills/smithers-agent-runtime/SKILL.mdgit clone --depth 1 https://github.com/smithersai/smithersWrote 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/skills/smithersai/smithers/smithers-agent-runtime)<a href="https://agentmods.dev/skills/smithersai/smithers/smithers-agent-runtime"><img src="https://agentmods.dev/badge/skills/smithersai/smithers/smithers-agent-runtime/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/smithersai/smithers/smithers-agent-runtime"><img src="https://agentmods.dev/badge/skills/smithersai/smithers/smithers-agent-runtime.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00030 | $0.00658 |
| Opus 5.5 | $0.00012 | $0.00263 |
| Sonnet 5.5 | $0.00006 | $0.00132 |
| Haiku 4.5 | $0.00003 | $0.00066 |
Grade A, and why
smithers-agent-runtime 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 5d 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.
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.
What it actually says
Smithers agent runtime
Read the agent composition guide and relevant package tests before changing host wiring or agent boundaries. Agent.Service owns one cell loop. Cells reach capabilities through ctx.call to ordinary flows; AgentSession and AgentAction adapt that same loop. A foreign CLI agent implements Agent.Service rather than adding a second loop.
The host resolves credentialed seats through SeatResolver, supplies quota and budget policy, and declares bounded sandbox limits. The completion evaluator is a host decision: select it before opening resources and refuse missing configuration rather than silently using a scripted or unavailable judge in production. Whether a model reads a completion at all is the workflow's decision: the claim brake is off unless the flow passes a positive claimCap (Will, 2026-10-14, #3993); never re-arm it as a harness or host default.
Budget accounting is durable. An unreadable or unwritable ledger is unknown spend and fails closed. Budget.check is advisory; a custom provider boundary needs scoped reserve and record around the real call. Check authorization before opening a journaled activity so a later grant can unblock a parked action. Keep the flow's declared capabilities and effects consistent with the bound implementation; verify recovery and refusal behavior in tests, not only a fresh run.
For model changes, read the model guide: canonical requests exclude credentials, which resolve just before transport. Ambient provider keys do not enable live tests; SMITHERS_LIVE_MODEL_TESTS=1 opts into quota-consuming calls. For command tools, read the Std guide: Bash hermetic is a lexical check, not OS confinement, and truncated output must stay visibly truncated. For scheduled or webhook launches, read the triggers guide: retryable claims require a durable idempotencyKey dedupe at RunnerService.start. For scoring receipts, read the scorers guide: a completed batch can contain inconclusive observations, and store failures are warnings; verify persistence by readback before claiming a grade was saved.
For host capability composition, read the kernel guide. Replace the ordinary host service tags with their guarded implementations in place; a parallel protected service lets dependencies bypass the guard. Path is intentionally unchecked, so a path service alone is no confinement boundary.
For untrusted execution, read the sandbox provider guide and choose a provider whose documented boundary meets the threat model. The package itself adds no isolation; DirectorySandbox is a scratch directory with process lifecycle, not filesystem, user, or network confinement. Verify provider cancellation with its conformance suite.
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.
- 5d ago Changed 9be053e8432c
- 26d ago First seen · 19 lines · 30 tokens per session scan A f4e90cf9f3b1
smithers-agent-runtime is a skill published in the GitHub repository smithersai/smithers (439 stars, last pushed today), licensed MIT. It adds 30 tokens to every session and 658 once invoked, about $0.0001 per session on Opus 5.5. 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 2026-09-25.
Other skills, from other repositories
paperclip
Use for Paperclip-managed tasks and heartbeats: reading task context, delivering task documents or files, updating completion or blockers, coordinating or delegating work, and following company governance. Includes control plane API operations for assignments, comments, approvals, and routines.
agentmail
Use your assigned AgentMail inbox to read email tasks, explicitly send or reply, and check delivery. Provided automatically by your inbox assignment.
cw-gates
Use before claiming any Codewhale change is done, green, or ready to land: the focused-to-broad verification ladder, the budget checks CI enforces, and the rules for what counts as a passing test.
refocusing
Use when a long-running agent may have lost the product outcome, when context was compacted, when work has crossed a major boundary, or when a fresh path-based trace is needed before the next consequential action. Not for fresh-session orientation, waking an idle seat, or checkpoints.
oversight-team
Use when you are a seat on the oversight pod (a standing monitor-mode rig that keeps OTHER rigs healthy), configuring or running the drift detectors, or choosing whether to intervene vs escalate. Covers the pull-not-poll posture, the v0 detectors (premature-park, process-drift, off-task, token-burn), the intervention…
ppt-review
Read this skill before the first deck or output quality review, so its criteria are available when the rendered images arrive. Apply the criteria to current officereview structural output and relevant rendered images. Do not use source code, a tool success flag, or the first-slide preview as a substitute for deck…