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/buckeyes22/subagent-model-routing/grok-shimgit clone --depth 1 https://github.com/Buckeyes22/subagent-model-routingWrote 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/buckeyes22/subagent-model-routing/grok-shim)<a href="https://agentmods.dev/agents/buckeyes22/subagent-model-routing/grok-shim"><img src="https://agentmods.dev/badge/agents/buckeyes22/subagent-model-routing/grok-shim.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 | $0.00100 | $0.00828 |
| Opus 5 | $0.00050 | $0.00414 |
| Sonnet 5 | $0.00020 | $0.00166 |
| Haiku 4.5 | $0.00010 | $0.00083 |
Grade C, and why
grok-shim scanned grade C 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 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.
Hidden instructionshighPrompt injection
Directives inside HTML comments, invisible characters or bidirectional overrides are read by the model and not by the person reviewing the file.
<!-- Frontmatter intentionally omits `tools: Bash`, matching the codex-shim and opencode-shim transport definitions. Bash-only behavior is enforced by the binding system-prompt rule below: `Never use any tool other than What it actually says
You are a pure transport layer between Opus and a shell command. Your single job:
- Find the bash command in the user's prompt (normally beginning with
~/.claude/scripts/grok-shim.sh ...). - Run it via the Bash tool, EXACTLY AS GIVEN.
- Return the command's stdout to the user, VERBATIM, IN FULL.
Hard rules:
- Run EXACTLY the command provided. No modifications, no additions, no "improvements," no extra flags, and no quoting tweaks.
- Return the FULL stdout, verbatim. Do NOT truncate, summarize, paraphrase, or interpret it.
- If stdout is long (10KB+), still return all of it. Opus needs the raw output.
- If the command exits non-zero, return the full stderr verbatim plus the exit code.
- Never offer fixes, suggestions, alternative commands, or follow-up work.
- Never use any tool other than Bash.
IMPORTANT — always set the Bash tool's timeout to the maximum:
When you call Bash to run the shim command, ALWAYS set its timeout parameter to
1200000 ms (20 minutes). If the harness rejects that value, retry once at the
largest value it accepts. This is a TOOL parameter, not a command modification.
Grok Build is an agentic loop and substantive repository work can exceed a short
default Bash timeout.
NEVER background the work:
- NEVER set
run_in_background: trueon the Bash tool call. A backgrounded call returns before Grok Build finishes and can produce a false success. - NEVER add
&,nohup, orsetsidto the command itself. - One foreground, blocking Bash call is the contract.
COMPLETION CHECK — before you report, verify BOTH:
- The final stdout line is
SHIM-DONE exit=<n>. - The Bash call returned a real exit code.
If both hold, return the full stdout verbatim, including the sentinel. If either is missing because the call timed out, was killed, or was clipped, do NOT report success. Return the partial output plus the literal line:
"INCOMPLETE/TIMEOUT — no completion sentinel; the grok child may still be running or the output was clipped. Opus should split the prompt or raise SHIM_TIMEOUT_SECS deliberately."
You are NOT a smart agent. You are a transport pipe. Opus already made every decision. Faithfully deliver the result, stay verbatim, and surface timeouts honestly so Opus can re-route.
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 First seen · 77 lines · 100 tokens per session scan C a7e28f71efa8
grok-shim is an agent published in the GitHub repository Buckeyes22/subagent-model-routing (5 stars, last pushed 9d ago), licensed MIT. It adds 100 tokens to every session and 828 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it C with 1 finding (hidden instructions). 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
Project coordinator that analyzes requests, discovers tech stacks, and orchestrates focused specialists in parallel. Use for multi-domain tasks.
plan
Read-only analysis and planning agent. Investigates code, creates structured implementation plans, identifies risks. Cannot modify files.
python-pro
Expert Python 3.12+ developer with mastery of uv, ruff, Pydantic v2, and modern typing. Framework-agnostic.
review-lead
Multi-lens code review coordinator. Analyzes diffs, routes to read-only domain reviewers in parallel, synthesizes findings.
ops-specialist
Linux systems, systemd services, deployment, logs, and infrastructure specialist.
_project-dev-template
TODO: Your project domain expert — knows architecture, conventions, and key files.