cxcc-subagent

cxcc-subagent is a skill for Claude Code from boldprojekte/franke_skills. It costs 40 tokens per session (1,685 once invoked), scanned A, original, MIT.

A supervisor for detached coding agents—separate Codex CLI, Claude Code, or Grok sessions running outside the current conversation.

In plain words
What is it for?
Use it for implementation, refactoring, migrations, bug fixes, test writing, codebase exploration, interface work, or end-to-end checks.
Why use it?
It lets you delegate work while tracking progress, handling stalled sessions, and receiving questions that need attention.

Skill for Claude Code

Written for Claude Code: shipped in a Claude Code plugin. Also seen: mentions subagents; mentions Claude Code; mentions Codex.

Part of the franke-skills plugin — 1 skill shipped together

Good fit Use it for implementation, refactoring, migrations, bug fixes, test writing, codebase exploration, interface work, or end-to-end checks.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/boldprojekte/franke_skills/cxcc-subagent
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 boldprojekte/franke_skills --skill cxcc-subagent
Clone the repo
git clone --depth 1 https://github.com/boldprojekte/franke_skills

Made for: Claude Code.

Or install franke-skills, the plugin that ships this one along with the rest of its 1 skill.

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 cxcc-subagent

README.md
[![agentmods](https://agentmods.dev/badge/skills/boldprojekte/franke_skills/cxcc-subagent/github.svg)](https://agentmods.dev/skills/boldprojekte/franke_skills/cxcc-subagent)
Your own site
<a href="https://agentmods.dev/skills/boldprojekte/franke_skills/cxcc-subagent"><img src="https://agentmods.dev/badge/skills/boldprojekte/franke_skills/cxcc-subagent/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 cxcc-subagent

Your own site · 80×15
<a href="https://agentmods.dev/skills/boldprojekte/franke_skills/cxcc-subagent"><img src="https://agentmods.dev/badge/skills/boldprojekte/franke_skills/cxcc-subagent.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 40 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,685 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 $0.00040 $0.01685
Opus 5 $0.00020 $0.00843
Sonnet 5 $0.00008 $0.00337
Haiku 4.5 $0.00004 $0.00169

Measured 2d ago against content hash b2a4a519e9e8, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-10, from the pricing page.

Security

Grade A, and why

cxcc-subagent 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 2d ago.

The scan reads SKILL.md. This mod also ships 5 executable files (scripts/cdx_core.py, scripts/cdx_version.py, scripts/cdx.py, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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.

skills/engineering/cxcc-subagent/SKILL.md · 72 lines

How it starts

The opening of the file, as written. The whole thing — 72 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CXCC Subagent

cdx supervises detached workers. You own the work order, decisions, review and final verification; workers execute bounded tasks with their own context.

Start

The CLI is scripts/cdx.py inside this skill directory and needs Python 3.10+. Invoke it by its absolute path. Run it without arguments to see this session's tasks and discover commands. Command-specific --help owns flags, defaults, output formats and examples.

Owner is derived from the harness session ID. If the home view shows a filesystem path as owner, set CDX_OWNER to a stable, session-specific value on every call to isolate parallel chats in the same directory.

Delegate and collect

  1. Bound the work. Give each worker a goal, repository and relevant paths, constraints, expected proof and output shape. Include decisions from this conversation that the worker cannot find in the repo. Reference existing material by path. Concurrent writers need separate worktrees or disjoint file ownership.
  2. Spawn. Pass a role file followed by the work order using repeatable -f. The returned task name identifies the worker; use it exactly. Continue independent work after dispatch.
  3. Observe. Use wait while workers run. If the harness yields a live process handle, poll that same process until the response arrives. Use watch --json instead only when the harness can deliver background lines as agent notifications; arm one watcher for the session. For watcher setup, cross-session recovery or backend-specific resume limits, read runtime.md.
  4. Act. Read state: working needs time; awaiting_reply needs an answer through send; done needs collection through result. For failed, stalled or killed, inspect peek and choose a corrected follow-up or a fresh worker. A stall_suspect flag means the worker is still running: inspect once and judge whether to wait or redirect. Use send --now for deliberate interruption.
  5. Verify. Read the result and actual repository diff. Credit concrete worker test evidence when it matches the change; re-run targeted checks for missing or contradictory evidence. After integration, run the full required gate once and any credited checks it does not cover. Finish when every task is accounted for and the integrated outcome is proven. Clean terminal tasks after collecting their results.

Read the full file on GitHub · 72 lines

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. 2d ago Changed · -81 lines · -98 tokens per session b2a4a519e9e8
  2. 10d ago First seen · 153 lines · 138 tokens per session scan A 9e64027394ec

Subscribe to this mod's changes

cxcc-subagent is a skill published in the GitHub repository boldprojekte/franke_skills (7 stars, last pushed 3d ago), licensed MIT. It adds 40 tokens to every session and 1,685 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 2026-08-31.

Related

Other skills, from other repositories