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.
npx skills add ComputerSaysNull/claude-delegate-local-mcp --skill write-delegate-agentgit clone --depth 1 https://github.com/ComputerSaysNull/claude-delegate-local-mcpWrote 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/skills/computersaysnull/claude-delegate-local-mcp/write-delegate-agent)<a href="https://agentmods.dev/skills/computersaysnull/claude-delegate-local-mcp/write-delegate-agent"><img src="https://agentmods.dev/badge/skills/computersaysnull/claude-delegate-local-mcp/write-delegate-agent/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/skills/computersaysnull/claude-delegate-local-mcp/write-delegate-agent"><img src="https://agentmods.dev/badge/skills/computersaysnull/claude-delegate-local-mcp/write-delegate-agent.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 | $0.00048 | $0.01210 |
| Opus 5 | $0.00024 | $0.00605 |
| Sonnet 5 | $0.00010 | $0.00242 |
| Haiku 4.5 | $0.00005 | $0.00121 |
Grade A, and why
write-delegate-agent scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 87 lines — stays where its author put it; the contents beside it link to each section on GitHub.
An agent file is a role this server can run: frontmatter that sets the dispatch, and a body
that instructs the model. delegate_to_agent and delegate_to_agent_readonly take its
name.
This file is itself a valid agent file in the format it describes, so it doubles as the worked example.
Where the file goes
Three locations, first match wins, so a project's own agent shadows a personal one of the same name:
<project>/.claude/agents/<name>.md<project>/.claude/skills/<name>/SKILL.md<agents_dir>/<name>.md— the personal tier, set byDELEGATE_AGENTS_DIR
The filename supplies the name — <name>.md, or the skill directory for a SKILL.md.
A name must match ^[A-Za-z0-9_-]+$: it is a name, not a path, so it cannot traverse.
Frontmatter
---
name: test-writer
description: Writes and extends pytest tests for a changed module.
model: deepseek-v4-flash
effort: low
max_turns: 20
max_tokens: 32768
keep_tool_results: 6
allowed_tools: [read_file, write_file, run_bash]
network: false
---
You write tests for existing code. Read the module, then write tests that would have
caught the bug described in the task. Run them. Iterate until they pass.
Report the outcome of the final test run.
A frontmatter block is required — a --- line, a closing ---, and something between
them: a file with no block, or whose --- lines are adjacent, is refused. No individual
field is required, so a block of one blank line yields an all-default agent.
Every key must be one of these ten. An unknown key is refused, not ignored, because a typo that costs you a setting in silence is the failure this format was rewritten to prevent:
| Key | Value | Notes |
|---|---|---|
name |
the agent's name | Optional, but if present it must equal the filename, or the file is refused — and a bare name: counts as present, because a key whose value went missing is a setting that silently does nothing |
description |
one line | What list_agents reports, so a caller can choose without opening the file |
model |
a model-registry key | Binds the dispatch |
effort |
off, low, high, max |
Refused loudly if misspelt. Not medium — the backend has no such level. Not inherit — that is what a caller passes to defer to this file |
max_turns |
integer ≥ 1 | Above the server's hard cap the file is refused, not clamped: a caller's number is transient, a file is committed and trusted |
max_tokens |
integer ≥ 1 | Reply budget; the model's own cap still applies afterwards |
keep_tool_results |
integer ≥ 0 | How many recent tool results survive history eviction. 0 evicts each one as the next turn starts |
allowed_tools |
[a, b] |
From read_file, search_files, read_git, run_bash, write_file, edit_file. Naming an unimplemented tool is refused. Brackets are the format: bare a, b is refused, and so is a - block list |
network |
true / false |
Default false. An operator grant: the agent must also be listed in DELEGATE_AGENT_NETWORK_ALLOWED and found in the personal tier |
extra_binds |
[a, b] |
Extra directories visible in the sandbox, within operator-set roots. Read-only, and the denylist still reaches inside. Bracketed like allowed_tools, and refused the same way |
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.
- 3d ago Changed e5118c7b8b18
- 4d ago Changed a667f64a5f74
- 5d ago First seen · 87 lines · 48 tokens per session scan A 1c0c78ed03ed
write-delegate-agent is a skill published in the GitHub repository ComputerSaysNull/claude-delegate-local-mcp (0 stars, last pushed today), licensed MIT. It adds 48 tokens to every session and 1,210 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 2026-09-11.
Other skills, from other repositories
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
local-ai-agents
Build local-first AI agents that run entirely on a developer workstation with Microsoft Foundry Local and Qwen function-calling models. Covers Small Language Models (SLMs), the OpenAI-compatible local endpoint, sandboxed local tools, local RAG with Chroma, local MCP servers, hybrid cloud/local routing, and the…
next-cache-components-adoption
Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…
next-cache-components-optimizer
Drive a Next.js route to instant navigation by setting up an agentic loop, under Cache Components / PPR, on initial load (hard navigation) and client-side navigation (soft navigation). Encode the goal as a failing @next/playwright instant() e2e and work it to green, one verified route at a time; the shipped test then…
next-partial-prefetching-adoption
Turn on Partial Prefetching in a Next.js app and work through the insights it surfaces. Use when the user wants to enable or adopt Partial Prefetching, flip the partialPrefetching flag, opt routes in with export const prefetch = 'partial', audit Link prefetch={true} behavior, preserve existing prefetched UI with…
chronicle
Analyze Copilot session history for standup reports, usage tips, session search, and session reindexing. Use when the user asks for a standup, daily summary, usage tips, workflow recommendations, wants to search or find past sessions by keyword/file/PR, wants to reindex their session store, or asks about deleting…