agentmods.dev

programmatic-agent-runs

programmatic-agent-runs is a skill for Cursor from IgorGanapolsky/ThumbGate. It costs 30 tokens per session (456 once invoked), scanned A, original, MIT.

Instructions for supervising coding-agent runs made through the Cursor SDK or similar local, cloud, and self-hosted workers.

In plain words
What is it for?
Recalling prior lessons, checking safety rules, choosing a runtime, limiting each agent’s scope, recording run details, and requiring verification before release actions.
Why use it?
They add checks before an unattended agent can edit files, create branches, open pull requests, publish packages, or deploy changes.

Skill for Cursor

Written for Cursor: shipped in a Cursor plugin. Also seen: mentions subagents.

Part of the thumbgate plugin — 5 skills, 3 commands, 2 agents shipped together

not rated 29repo yesterday A scan SkillSpector: pass 30 tokens original MIT

Good fit Recalling prior lessons, checking safety rules, choosing a runtime, limiting each agent’s scope, recording run details, and requiring verification before release actions.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/igorganapolsky/thumbgate/programmatic-agent-runs
View source ↗ IgorGanapolsky/ThumbGate thumbgate.ai
Install

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.

Any agent
npx skills add IgorGanapolsky/ThumbGate --skill programmatic-agent-runs
Clone the repo
git clone --depth 1 https://github.com/IgorGanapolsky/ThumbGate

Made for: Cursor.

Or install thumbgate, the plugin that ships this one along with the rest of its 5 skills, 3 commands, 2 agents.

Wrote 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.

agentmods badge for programmatic-agent-runs

README.md
[![agentmods](https://agentmods.dev/badge/skills/igorganapolsky/thumbgate/programmatic-agent-runs/github.svg)](https://agentmods.dev/skills/igorganapolsky/thumbgate/programmatic-agent-runs)
Your own site
<a href="https://agentmods.dev/skills/igorganapolsky/thumbgate/programmatic-agent-runs"><img src="https://agentmods.dev/badge/skills/igorganapolsky/thumbgate/programmatic-agent-runs/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.

agentmods 80×15 button for programmatic-agent-runs

Your own site · 80×15
<a href="https://agentmods.dev/skills/igorganapolsky/thumbgate/programmatic-agent-runs"><img src="https://agentmods.dev/badge/skills/igorganapolsky/thumbgate/programmatic-agent-runs.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 30 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 456 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe. Third-party audits
  • NVIDIA SkillSpector pass 7 Sept 2026
How audits are shown
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00030 $0.00456
Opus 5.5 $0.00012 $0.00182
Sonnet 5.5 $0.00006 $0.00091
Haiku 4.5 $0.00003 $0.00046

Measured 1mo ago against content hash b1f85b1b5529, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

programmatic-agent-runs 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 1mo 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.

plugins/cursor-marketplace/skills/programmatic-agent-runs/SKILL.md · 43 lines

What it actually says

Programmatic Agent Runs

Use this skill before launching a coding agent through the Cursor SDK, Cursor cloud agents, self-hosted workers, or subagents.

When to use

  • A task starts from code, CI, a backend service, a kanban automation, or any unattended workflow
  • The run can edit files, push a branch, create a PR, publish a package, deploy, or call shell tools
  • The run delegates work to subagents with different prompts, models, or scopes

Launch checklist

  1. Recall lessons with search_lessons for the repo, branch, files, and intended action.
  2. Check active gates with prevention_rules before enabling writes or autoCreatePR.
  3. Use the narrowest runtime: local for developer-in-the-loop work, cloud for isolated async work, self-hosted for private network/data requirements.
  4. Give each subagent a bounded file or responsibility scope.
  5. Persist run metadata: runtime, repo URL, starting ref, run id, agent id, branch, PR URL, and linked ThumbGate evidence.
  6. Require verification output before merge, publish, deploy, or customer-facing claims.

Cursor SDK guardrails

  • Attach ThumbGate MCP through the run environment so the agent can retrieve lessons and gates.
  • Prefer a clean starting ref and a disposable branch for cloud VM runs.
  • Do not set autoCreatePR for destructive, billing, release, or production tasks unless the gate check is clean.
  • If a cloud run finishes while the developer is offline, inspect the run transcript and PR diff before treating the result as accepted.
  • Capture thumbs_down feedback when a run violates scope, skips proof, repeats a known mistake, or opens a noisy PR.

Output

For each governed run, report:

  • Runtime: local, cloud, or self-hosted
  • Scope: files, modules, or workflow the agent may touch
  • Gates checked and result
  • Verification evidence
  • Branch or PR URL
  • Any feedback captured for future prevention
Changes

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.

  1. 1mo ago First seen · 43 lines · 30 tokens per session scan A b1f85b1b5529

Subscribe to this mod's changes

programmatic-agent-runs is a skill published in the GitHub repository IgorGanapolsky/ThumbGate (29 stars, last pushed yesterday), licensed MIT. It adds 30 tokens to every session and 456 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 2026-08-30.

AI SafetyBuilding agentsGuardrailsagent-reliabilityai-agentsdeveloper-toolsthompson-sampling
Related

Other skills, from other repositories

llamaguard

Meta's 7-8B specialized moderation model for LLM input/output filtering. 6 safety categories - violence/hate, sexual content, weapons, substances, self-harm, criminal planning. 94-95% accuracy. Deploy with vLLM, HuggingFace, Sagemaker. Integrates with NeMo Guardrails.

davila7/claude-code-templates · 74 tokens

llamaguard

Meta's 7-8B specialized moderation model for LLM input/output filtering. 6 safety categories - violence/hate, sexual content, weapons, substances, self-harm, criminal planning. 94-95% accuracy. Deploy with vLLM, HuggingFace, Sagemaker. Integrates with NeMo Guardrails.

Orchestra-Research/AI-Research-SKILLs · 74 tokens

llamaguard

Meta's 7-8B specialized moderation model for LLM input/output filtering. 6 safety categories - violence/hate, sexual content, weapons, substances, self-harm, criminal planning. 94-95% accuracy. Deploy with vLLM, HuggingFace, Sagemaker. Integrates with NeMo Guardrails.

synthetic-sciences/openscience · 74 tokens

agentguard

Runtime guardrails for AI coding agents. Recorded-budget preflight, loop, retry, and timeout checks in instrumented Python. Zero dependencies, local-first, MIT licensed.

bmdhodl/agent47 · 37 tokens

llamaguard

Meta's 7-8B specialized moderation model for LLM input/output filtering. 6 safety categories - violence/hate, sexual content, weapons, substances, self-harm, criminal planning. 94-95% accuracy. Deploy with vLLM, HuggingFace, Sagemaker. Integrates with NeMo Guardrails.

OpenLAIR/dr-claw · 74 tokens

vv-gate

Use when an agent, harness, or CI pipeline produces a scientific/numeric prediction and you need an INDEPENDENT third-party adjudication of whether that claim is trustworthy before acting on it. Also use when you need to turn "the agent said it validated itself" into a blocking gate with a real exit code. Covers…

lm203688/swarmlabs-engine-kit · 107 tokens