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/MadAppGang/magusWrote 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/commands/madappgang/magus/delegate)<a href="https://agentmods.dev/commands/madappgang/magus/delegate"><img src="https://agentmods.dev/badge/commands/madappgang/magus/delegate/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/commands/madappgang/magus/delegate"><img src="https://agentmods.dev/badge/commands/madappgang/magus/delegate.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 | $0.00111 | $0.03385 |
| Opus 5.5 | $0.00044 | $0.01354 |
| Sonnet 5.5 | $0.00022 | $0.00677 |
| Haiku 4.5 | $0.00011 | $0.00338 |
Grade A, and why
delegate 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 today.
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.
How it starts
The opening of the file, as written. The whole thing — 240 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Rules
- CHILD MODE: YOU ARE THE DELEGATE — check for child mode as
claudish:claudish-usage→ "Inside a claudish session (child mode)" describes. In child mode, claudish allows no session below this one. Say so in one line, then do the task yourself: the rules and phases below describe launching a session, so skip them, and run any/plugin:commandthe user named yourself. If claudish has no tools at all, that is not child mode: report it as a failure under NO AUTO-RECOVERY. - NO AUTO-RECOVERY — on failure, report verbatim and stop. No retries, no substitution. Resolution may default when nothing was named; recovery never substitutes a named model. Picking a model the user did not specify is a documented, announced default (Step 1c.4). Swapping a model the user did specify, after it failed, is substitution and stays forbidden.
- NO PRE-SOLVING — do not read project files before launching. The external model investigates itself.
- FORWARD A QUESTION TO USER — Phase 2 starts the session with a
prompt, so with claudish 10.4.0 or later it never waits for input. If a question arrives anyway (input_required, orneeds-input … next: send_input), use AskUserQuestion whenever it is available. Never auto-answer. Where it is unavailable (claude -p), see Phase 3 — cancel the session and report the question rather than hanging on an answer that cannot arrive. - NEVER add provider prefixes — no "openai/", "google/", "mm@", "or@". Claudish resolves internally.
Phase 1: Parse and Resolve
Step 1a — Load alias table: Follow the claudish:claudish-usage skill → "Model Alias Resolution" procedure to build ALIAS_TABLE from the live catalog (list_models) + .claude/multimodel-team.json customAliases.
Step 1b — Parse arguments left-to-right:
- First token with no
/or--prefix → MODEL_ARG - Token starting with
/and containing:→ resolve it before deciding what it is:- matches a command or skill → EXPLICIT_COMMAND (inlined into the prompt)
- matches an agent (
plugins/<ns>/agents/*.mdfrontmattername:) → EXPLICIT_AGENT - matches neither → say so and ask; do not inline an unresolvable token
- Remaining tokens → TASK_DESCRIPTION
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.
- today Changed · +6 lines 85458de4db7d
- 10d ago Changed · +14 lines 7c65cc183d3e
- 22d ago Changed c3e26ce8b50a
- 1mo ago First seen · 220 lines · 111 tokens per session scan A b0665ec20acd
delegate is a command published in the GitHub repository MadAppGang/magus (13 stars, last pushed today), licensed MIT. It adds 111 tokens to every session and 3,385 once invoked, about $0.0004 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 2026-09-04.
Other commands, from other repositories
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
specify
Create or update the feature specification from a natural language feature description.
converge
Assess the current codebase against the feature's spec, plan, and tasks, then append any remaining unbuilt work as new tasks to tasks.md so implement can complete it.
implement
Execute the implementation plan by processing and executing all tasks defined in tasks.md.
speckit.bug.fix
Apply the remediation from a bug assessment and record what was changed.