safe-autonomous-run

safe-autonomous-run is a skill for Claude Code, Codex from icohangar-ops/agent-conductor. It costs 52 tokens per session (396 once invoked), scanned A, original, MIT.

A controlled loop for running an AI agent with spending limits, cost previews, approval pauses, and an emergency stop. It requires authorization before model calls and before high-impact shell commands.

In plain words
What is it for?
It is for supervising bounded agent runs, approving sensitive actions, enforcing budgets, and stopping runaway work.
Why use it?
It prevents an autonomous run from spending indefinitely, looping without progress, or executing risky commands without operator approval. A kill switch stops further authorized actions.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Good fit It is for supervising bounded agent runs, approving sensitive actions, enforcing budgets, and stopping runaway work.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/icohangar-ops/agent-conductor/safe-autonomous-run
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 icohangar-ops/agent-conductor --skill safe-autonomous-run
Clone the repo
git clone --depth 1 https://github.com/icohangar-ops/agent-conductor

Made for: Claude Code, Codex.

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 safe-autonomous-run

README.md
[![agentmods](https://agentmods.dev/badge/skills/icohangar-ops/agent-conductor/safe-autonomous-run/github.svg)](https://agentmods.dev/skills/icohangar-ops/agent-conductor/safe-autonomous-run)
Your own site
<a href="https://agentmods.dev/skills/icohangar-ops/agent-conductor/safe-autonomous-run"><img src="https://agentmods.dev/badge/skills/icohangar-ops/agent-conductor/safe-autonomous-run/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 safe-autonomous-run

Your own site · 80×15
<a href="https://agentmods.dev/skills/icohangar-ops/agent-conductor/safe-autonomous-run"><img src="https://agentmods.dev/badge/skills/icohangar-ops/agent-conductor/safe-autonomous-run.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 52 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 396 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.
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 $0.00052 $0.00396
Opus 5 $0.00026 $0.00198
Sonnet 5 $0.00010 $0.00079
Haiku 4.5 $0.00005 $0.00040

Measured 3d ago against content hash 5a48b71398bb, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

safe-autonomous-run 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 3d 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.

examples/safe-autonomous-run/.conductor/skills/safe-autonomous-run/SKILL.md · 35 lines

What it actually says

Safe autonomous run

Drive a Conductor-bounded loop. The controller — not the model — decides when spend stops.

Steps

  1. run_preflight with the system rules, history, tool schemas, and user prompt. Treat the USD figure as approximate (ceil(chars/4)).
  2. run_begin with path pointing at this example (or pass explicit runUsd / tenantDayUsd). A missing mandate is a HALT.
  3. For each model turn: run_authorize (kind=model) → perform the call only on PASS → run_commit with actual tokens/USD if the provider reported them.
  4. For Bash (high-impact): expect APPROVE_REQUIRED. An operator runs decision_gate / run_approve with solvable, scoped, valid, worth_it before the next authorize.
  5. On HALT (run_ceiling, tenant_day_ceiling, stuck_loop, kill_switch, hard_stop) stop. Do not invent a new run id to bypass the ceiling.
  6. run_kill at any time aborts further authorizes without asking the model.

Constraints

  • Never call run_commit without a clearance id from run_authorize.
  • Never raise maxModelClass after a HALT to "finish the task".
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. 3d ago First seen · 35 lines · 52 tokens per session scan A 5a48b71398bb

Subscribe to this mod's changes

safe-autonomous-run is a skill published in the GitHub repository icohangar-ops/agent-conductor (0 stars, last pushed 3d ago), licensed MIT. It adds 52 tokens to every session and 396 once invoked, about $0.0003 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 2026-09-07.

Related

Other skills, from other repositories

repo-context

Use repo-context-mcp tools to map, search, and pack repository context before large edits or PR review.

nduc99911/repo-context-mcp · 25 tokens

social-scrape

Collect public posts and discussion from X, LinkedIn and Hacker News through a local signed-out browser, then report them in chat or write them to a document. Use when the user wants to scrape, monitor or research specific accounts, topics, hashtags or discussions on social platforms.

alijancb/subio-mcp · 60 tokens

MCP Integration Assistant

Helps design and implement Model Context Protocol (MCP) server integrations for AI agents.

Notysoty/openagentskills · 23 tokens

MCP Server Builder

Step-by-step guidance for creating a new MCP server with FastMCP or the TypeScript SDK — tool definitions, resource handlers, error responses, and usage examples.

Notysoty/openagentskills · 38 tokens

run-user-interviews

Run real user interviews (voice or text) through Usercall and return themes with verbatim quotes. Use when the user wants qualitative feedback, onboarding research, usability tests, prototype review, or to talk to real users instead of guessing.

junetic/usercall-mcp · 52 tokens

mapping-dependencies

Maps task dependencies in an Astravue project, finds the critical path and blocked work, and detects circular dependencies before any rescheduling. Use when the user asks about dependencies, blockers, what's blocking what, the critical path, or wants to understand or fix the sequencing of a project.

AstravueOrg/astravue-mcp-server · 61 tokens