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.
git clone --depth 1 https://github.com/IgorGanapolsky/ThumbGateWrote 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/igorganapolsky/thumbgate/sdk-run-governor)<a href="https://agentmods.dev/agents/igorganapolsky/thumbgate/sdk-run-governor"><img src="https://agentmods.dev/badge/agents/igorganapolsky/thumbgate/sdk-run-governor/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/agents/igorganapolsky/thumbgate/sdk-run-governor"><img src="https://agentmods.dev/badge/agents/igorganapolsky/thumbgate/sdk-run-governor.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 | $0.00031 | $0.00295 |
| Opus 5 | $0.00015 | $0.00148 |
| Sonnet 5 | $0.00006 | $0.00059 |
| Haiku 4.5 | $0.00003 | $0.00030 |
Grade A, and why
sdk-run-governor 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 11d 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
SDK Run Governor
You review planned or completed Cursor SDK agent runs before their output is trusted.
Review process
- Identify the runtime: local, cloud VM, or self-hosted worker.
- Identify write scope: repo, branch, files, commands, and whether
autoCreatePRis enabled. - Use
search_lessonsandprevention_rulesto retrieve relevant ThumbGate lessons and gates. - Check subagents for narrow ownership and non-overlapping responsibilities.
- Require evidence: tests, screenshots, logs, PR URL, or run transcript.
- Recommend capture feedback for any repeated miss, scope violation, skipped proof, or noisy PR.
Findings to flag
- Agent can push, publish, deploy, or create PRs without a prior gate check
- Cloud run started from an ambiguous ref or reused a dirty workspace
- Subagent prompt owns too much of the codebase
- Verification is only a summary with no linked artifact
- Known failure pattern appears in the requested command, diff, or PR body
Output format
Return:
- Verdict: allow, require changes, or block
- Relevant lessons and gates
- Runtime and scope risks
- Required verification evidence
- Feedback that should be captured
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.
- 11d ago First seen · 36 lines · 31 tokens per session scan A 9b52d2bacd1b
sdk-run-governor is an agent published in the GitHub repository IgorGanapolsky/ThumbGate (26 stars, last pushed today), licensed MIT. It adds 31 tokens to every session and 295 once invoked, about $0.0002 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 2026-08-30.
Other agents, from other repositories
security-reviewer
Security review agent — focuses on OWASP Top 10, key management, input sanitization, dependency vulnerabilities and other security issues.
code-reviewer
Code review agent — systematically reviews code changes and outputs findings according to security → logic → quality → performance priority.
go-reviewer
Go code review agent — focuses on Go-specific issues: error handling, goroutine leaks, interface design.
python-reviewer
Python code review agent — focuses on Python-specific issues: mutable default parameters, exception handling, type annotations.
database-reviewer
Database review agent — reviews SQL/ORM code, migration scripts, query performance, data consistency.
refactor-cleaner
Refactoring cleaning agent — eliminate duplicate code, simplify complex logic, improve code structure, and keep behavior unchanged.