Sponsor Suno AI Music arrow_forward
Plugin

Conjure

Intelligent delegation framework for routing tasks to external LLM services while retaining strategic oversight

Type
Plugin
GitHub stars
339
License
MIT
Repo last updated
Sep 24, 2026
Version
1.9.20
Author
Alex Thola

What Conjure is

Conjure is a plugin published in the athola/claude-night-market repository on GitHub, which has about 339 stars. The repository describes itself as: “23 Claude Code plugins: TDD enforcement hooks, git/PR workflows, spec-driven development, code review, project lifecycle, fix-from-error, maintenance automation, context optimization, research, and multi-LLM delegation. 186 skills, 128 commands, 54 agents.”

A plugin is a package that bundles skills, slash commands, subagents, hooks, and MCP connectors so they install together. Plugins are plain files with a manifest at .claude-plugin/plugin.json, and they work in both Claude Code and Claude Cowork.

Installing Conjure adds everything it ships in one step. Connectors inside a plugin still need to be connected separately, and hooks and subagents only run in Cowork and Claude Code, not in regular chat.

How to install Conjure

Claude Code

  1. Add the repository as a plugin marketplace: claude plugin marketplace add athola/claude-night-market
  2. Install the plugin: claude plugin install conjure@<marketplace-name>, using the marketplace name from the repository's .claude-plugin/marketplace.json.
  3. Restart the session if the new skills or commands don't appear straight away.

Claude Cowork

  1. Open Customize → Plugins and choose Add marketplace.
  2. Enter athola/claude-night-market (the owner/repo shorthand works for GitHub).
  3. Find Conjure in the list, click Install, then connect any connectors it needs from its Connectors tab.

New to extending Cowork? Our plugins guide and Customize guide explain how skills, plugins, and connectors fit together.

Inside the source file

An excerpt from plugins/conjure/.claude-plugin/plugin.json, shared under the repository's MIT license. Read the full file on GitHub.

Delegate tasks to external models from Claude Code. Delegate analysis, bulk work, and summarization to Gemini, Qwen, MiniMax, GLM, Meta Muse, Codex, OpenCode, or a Muse Glimmer running locally.

Track quotas, log usage, and suggest delegation for large tasks.

Installation

As a Claude Code plugin

/plugin install athola@claude-night-market
/status

Development setup

uv sync             # install deps
make install-hooks  # pre-commit hooks
make test           # lint + type + security checks

Requirements: Python 3.10+, uv.

Optional Dependencies

For accurate token counts, install tiktoken:

pip install tiktoken

Why delegation is on by default

It was opt-in until the provider chain landed, under [docs/inclusive-defaults.md][inc] TRUE-exception category 3: the external CLIs (gemini, qwen, mmx, muse, codex, opencode, claude, ollama) are separately installed and authenticated, and smart_delegate raised when none of them was. Defaulting on would have failed every task on an unconfigured machine.

The chain removed that failure. An unconfigured machine now costs one availability probe per provider and gets a providers_exhausted result, which the caller reads as "do this work yourself". The reasonable default the category asked for exists, and it is "try, then do it yourself".

Skill(conjure:provider-setup) is how an operator stops paying that cost. It prints which CLIs are installed and authenticated, explains each unhealthy one, installs the missing ones after confirmation, and stores the result at ~/.claude/hooks/delegation/provider-state.json:

python3 scripts/delegation_setup.py --available
python3 scripts/delegation_setup.py --doctor
python3 scripts/delegation_setup.py --install gemini

Two switches decline it, environment over file:

A disabled delegator probes and spawns nothing.

[inc]: ../../docs/inclusive-defaults.md

Usage

Quick Start

# Check delegation readiness
make delegate-verify

# Select best service for a task
make delegate-auto PROMPT="Summarize src" FILES="src/"

# Monitor limits and usage
make quota-status
make usage-report

Delegation Executor

# List services
uv run python scripts/delegation_executor.py --list-services

# Verify a service
uv run python scripts/delegation_executor.py --verify gemini

# Auto-select a service for the task
uv run python scripts/delegation_executor.py auto "Analyze this code" \
  --files src/

# Force a specific service
uv run python scripts/delegation_executor.py gemini "Summarize" \
  --files docs/*.md --model gemini-3-pro

Make Commands

# Development
make format          # ruff format + check --fix
make lint            # ruff check
make typecheck       # mypy + ty
make security-check  # bandit
make test            # lint + type + security bundle
make validate-all    # full validation including hooks
make clean           # remove caches/venv

# Delegation lifecycle
make delegate-status
make delegate-verify
make delegate-usage
make delegate-test
make delegate-gemini PROMPT="Analyze" FILES="src/main.py"
make delegate-qwen   PROMPT="Extract" FILES="src/**/*.py"
make delegate-minimax PROMPT="Summarize" FILES="src/**/*.py"
make delegate-glm     PROMPT="Review this design"
…

Quota & Usage Tools

# Quota tracker (Gemini)
uv run python tools/quota_tracker.py --status
uv run python tools/quota_tracker.py --estimate src/ docs/
uv run python tools/quota_tracker.py --validate-config

# Usage logger (Gemini)
uv run python tools/usage_logger.py --report
uv run python tools/usage_logger.py --validate
uv run python tools/usage_logger.py --status

In Claude Code

Use skills directly in chat:

Skill(conjure:delegation-core)
Skill(conjure:gemini-delegation)
Skill(conjure:qwen-delegation)
Skill(conjure:minimax-delegation)
Skill(conjure:glm-delegation)
Skill(conjure:muse-delegation)
Skill(conjure:codex-delegation)
Skill(conjure:opencode-delegation)

Hooks surface delegation suggestions for large tasks.

Commands

delegate-auto

Select the best external service based on requirements.

delegate-

Force a specific service with optional file globs and model hints. One target per registered provider: gemini, qwen, minimax, glm, muse, codex, opencode, glimmer.

delegate-setup / delegate-doctor / delegate-install

Report which provider CLIs are present and authenticated, name the command that fixes each unhealthy one, and install the missing ones after a confirmation showing the package, publisher, and source URL. Install commands come only from the vetted provenance map, so an unrecorded binary is refused rather than guessed at.

Before you install

  • Read the whole file first. Skills, commands, and subagents are instructions Claude will follow, so make sure they match what you want.
  • Check which tools, scripts, or MCP servers it uses. Local servers and scripts run with your permissions.
  • Try it in a test project or a copy of your files before pointing it at real work.
  • Pin the version you tested, and review changes before updating.
  • Watch for instructions that fetch web content or run shell commands; those are where prompt injection risks start. See our prompt injection guide.

FAQ

What is Conjure?

Conjure is a plugin for Claude Code and Claude Cowork from the athola/claude-night-market repository on GitHub. Intelligent delegation framework for routing tasks to external LLM services while retaining strategic oversight

How do I install Conjure in Claude Code?

Add the repository as a plugin marketplace: claude plugin marketplace add athola/claude-night-market Install the plugin: claude plugin install conjure@<marketplace-name>, using the marketplace name from the repository's .claude-plugin/marketplace.json. Restart the session if the new skills or commands don't appear straight away.

Can I use Conjure in Claude Cowork?

Open Customize → Plugins and choose Add marketplace. Enter athola/claude-night-market (the owner/repo shorthand works for GitHub). Find Conjure in the list, click Install, then connect any connectors it needs from its Connectors tab.

Is Conjure safe to install?

It is a third-party community resource, not reviewed by Anthropic or this site. Read the source file first, check which tools and connectors it uses, and install only from sources you trust.

Similar resources

Browse all skills, subagents, and plugins →

Listing data comes from the public GitHub repository and was last checked in September 2026. Excerpts are © their authors and shared under MIT. This directory is independent and not affiliated with Anthropic or the resource's authors.