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 agentmods add skills/modiqo/skillspec/skill-routernpx skills add modiqo/skillspec --skill skill-routergit clone --depth 1 https://github.com/modiqo/skillspecWrote 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/modiqo/skillspec/skill-router)<a href="https://agentmods.dev/skills/modiqo/skillspec/skill-router"><img src="https://agentmods.dev/badge/skills/modiqo/skillspec/skill-router.svg" alt="Measured on agentmods" 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 | $0.00839 |
| Opus 5 | $0.00026 | $0.00419 |
| Sonnet 5 | $0.00010 | $0.00168 |
| Haiku 4.5 | $0.00005 | $0.00084 |
Grade A, and why
skill-router 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 6d 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 — 73 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Skill Router
This skill is a thin native harness loader for the SkillSpec router. When router mode is enabled and the harness has been restarted, this router is the first hop for every user request in managed skill roots. All routed skills are explicit-only/manual-only, so check the router index before reading, searching for, or guessing at a domain skill.
The router does not execute task work. It decides whether a local skill should
be loaded, reports confidence and candidate paths, and asks for direct or
durable execution only after a use_skill decision when the task will use tools
and the user has not already chosen an execution mode.
Runtime Contract
-
For ordinary requests with prompt-hook context
first_hop_ready=true, use the fast path: do not read./skill.spec.ymland do not runskillspec router index status. -
Run one route query:
skillspec route --index <index-file-or-router-dir> --query '<user task>' --top 5 --json -
If route output says
decision: "use_skill"andselectedis non-null, read that skill'sSKILL.mdand follow it. -
If route output says
decision: "bypass", do not load any candidate skill; continue with normal agent behavior for the request. -
If route output says
decision: "ambiguous", do not silently choose a candidate. Ask only when the user explicitly requested skill selection; otherwise continue with normal agent behavior. -
If route output includes
execution_mode_direct_or_durablefor ause_skilldecision, ask the user whether to run direct or durable before tool-backed execution. -
Load and follow
./skill.spec.ymlonly for router lifecycle, repair, visibility, guard, index status, index refresh, or when guard context is missing or failed. -
If the user chooses durable execution, hand the selected skill and task context to
durable-executor. The router still only routes; durable-executor owns workspace evidence, rote execution policy, alignment, token stats, and final closure.
What ships with it
1 file 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.
- 6d ago First seen · 73 lines · 51 tokens per session scan A b6732e20b41a
skill-router is a skill published in the GitHub repository modiqo/skillspec (739 stars, last pushed 28d ago), licensed Apache-2.0. It adds 51 tokens to every session and 839 once invoked, about $0.0003 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 2026-08-30.
Other skills, from other repositories
ashfox
Create, edit, review, animate, and export game-ready low-poly assets in the ashfox web workbench through an AI agent. Use for low-poly modeling, deterministic textures and UVs, rigs, idle or motion animation, GLB, Bedrock, or GeckoLib exports, and Blockbench-free asset production.
ui-2026-cinematic
Cinematic 3D / scroll-driven / motion design skill for landing pages in 2026. Use when building marketing pages, SaaS sites, or product showcases that need scroll storytelling, 3D elements, micro-interactions, kinetic typography, or glassmorphism. Encodes modern best practices from Figma, Anthropic, Apple, Vercel.…
template-skill
A minimal template for creating a new FDE Consultant Skill (Apache-2.0). Use as starting point for any new skill.
production-autopsy
Start here. Audits a deployed ML or LLM or agent system that scores well on evaluation but fails, regresses, or behaves unexpectedly in production. Runs a reproducible root-cause "autopsy": frames the eval-to-deployment gap, reproduces the production failure, quantifies it by slice, tests confidence calibration under…
calibration-guard
Detects and quantifies confidently-wrong behavior: where a model, classifier, agent, or LLM judge is highly confident and incorrect, and where the coupling between confidence and correctness breaks down under distribution shift. Measures calibration on the in-distribution set and again on the production distribution…
tool-eval
Verifies the tools an agent or orchestrator depends on, by re-deriving a tool evaluation's real accuracy instead of trusting a single pass/fail score. Separates a formatting miss (a correct value scored wrong) from a real failure (a wrong value scored right), recomputes the expected answer from the raw inputs rather…