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 krzysztofsurdy/code-virtuoso --skill agent-designgit clone --depth 1 https://github.com/krzysztofsurdy/code-virtuosoWrote 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/krzysztofsurdy/code-virtuoso/agent-design)<a href="https://agentmods.dev/skills/krzysztofsurdy/code-virtuoso/agent-design"><img src="https://agentmods.dev/badge/skills/krzysztofsurdy/code-virtuoso/agent-design/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/krzysztofsurdy/code-virtuoso/agent-design"><img src="https://agentmods.dev/badge/skills/krzysztofsurdy/code-virtuoso/agent-design.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 | $0.00051 | $0.03161 |
| Opus 5.5 | $0.00020 | $0.01264 |
| Sonnet 5.5 | $0.00010 | $0.00632 |
| Haiku 4.5 | $0.00005 | $0.00316 |
Grade A, and why
agent-design 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 yesterday.
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 — 159 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent design
An LLM agent is a loop: the model picks a tool, the harness runs it, the result goes back into context, and the loop repeats until a stop condition fires. Most of the quality, cost, and safety of the system comes from the harness around that loop, not from the prompt. This skill holds the design decisions that hold across frameworks and model vendors. Library syntax changes monthly, so fetch it through docs-lookup instead of recalling it.
Core principles
| Principle | Meaning |
|---|---|
| Simplest thing that works | A single model call with good context beats a loop. A loop beats multi-agent. Add structure only when a measured failure demands it |
| Workflow before agent | If you can draw the steps before the run, write them as code. Let the model choose steps only when the path cannot be predicted |
| The harness is the product | Stop conditions, caps, permissions, and recovery decide whether the agent is safe to run unattended |
| Context is a budget | Reliability falls as context grows. Curate what the model sees at each step |
| Output is untrusted input | Tool arguments, structured output, and retrieved text can be wrong or hostile. Validate and authorize in code |
| Evals before architecture changes | Without a task set, you cannot tell whether a pattern, model, or framework helped |
Workflow or agent
Anthropic's definitions: a workflow runs LLMs and tools through predefined code paths, and an agent lets the LLM direct its own process and tool use (Building effective agents).
| Question | Workflow | Agent |
|---|---|---|
| Can you list the steps in advance? | Yes | No, the number of steps depends on what the tools return |
| Cost and latency | Predictable, lower | Variable, higher, with compounding-error risk |
| Testing | Test each step, then the chain | Test trajectories and outcomes, many trials per task |
| Failure mode | A step fails in a known place | The model wanders, loops, or stops early |
| Typical fit | Extraction, classification, report generation, support triage | Coding, open-ended research, multi-system operations |
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- yesterday First seen · 159 lines · 51 tokens per session scan A c4e10c163e62
agent-design is a skill published in the GitHub repository krzysztofsurdy/code-virtuoso (36 stars, last pushed today), licensed MIT. It adds 51 tokens to every session and 3,161 once invoked, about $0.0002 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 2026-10-22.
Other skills, from other repositories
hermes-diagnostic-review
Use when running a read-only diagnostic review of recent Hermes sessions to find recurring mistakes, failed tool calls, and repeated fixes, then propose suggestion-only improvements and reusable skills. Human-gated; never auto-applies.
instruction-eval
Change one condition an agent runs under (instruction text, the reference material instructions point at, MCP tools and permissions, hooks, skill files), then run the same prompts before and after, several times each, to see what actually changes. Produces an HTML report showing both arms' answers side by side. Use…
worth-building
Decide how elaborate to build something (simple vs. robust) and return a concrete, PoC-shaped proposal at that level — not over- or under-engineered, but right-sized. Use whenever someone is building something and the investment level is unclear: "quick or proper?", "MVP vs. real product", "is this over-engineering?"…
django-ticket-triage
Analyze a Django Trac ticket and produce a triage recommendation report — duplicate search, related PRs, forum threads, and the affected source code. Use when the user gives a Django ticket number, or asks whether a ticket is valid, a duplicate, or ready for a triage stage.
gh-issues
Fetch GitHub issues, spawn sub-agents to implement fixes and open PRs, then monitor and address PR review comments. Usage: /gh-issues [owner/repo] [--label bug] [--limit 5] [--milestone v1.0] [--assignee @me] [--fork user/repo] [--watch] [--interval 5] [--reviews-only] [--cron] [--dry-run] [--model glm-5]…
healthcheck
Host security hardening and risk-tolerance configuration for OpenClaw deployments. Use when a user asks for security audits, firewall/SSH/update hardening, risk posture, exposure review, OpenClaw cron scheduling for periodic checks, or version status checks on a machine running OpenClaw (laptop, workstation, Pi, VPS).