Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/KaydenClark/LLM_Workbenchnpx agentmods add skills/kaydenclark/llm_workbench/workbench-evaluationWrote 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/kaydenclark/llm_workbench/workbench-evaluation)<a href="https://agentmods.dev/skills/kaydenclark/llm_workbench/workbench-evaluation"><img src="https://agentmods.dev/badge/skills/kaydenclark/llm_workbench/workbench-evaluation/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/kaydenclark/llm_workbench/workbench-evaluation"><img src="https://agentmods.dev/badge/skills/kaydenclark/llm_workbench/workbench-evaluation.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 | $0.00000 | $0.02306 |
| Opus 5.5 | $0.00000 | $0.00922 |
| Sonnet 5.5 | $0.00000 | $0.00461 |
| Haiku 4.5 | $0.00000 | $0.00231 |
Grade A, and why
workbench-evaluation 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 12d 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 — 204 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Workbench evaluation
The evaluation and feedback operations this repository's maintainers run, moved here from the Runbook by the Contract Carrier Pointer-Brief Rewrite (S-004C TK-005K). Each section below is the procedure an operations index row in RUNBOOK.md points to, and it binds for that operation in this repository. The guardrail audit stays with the implement skill and the Runbook's Guardrail North-Star Audit section.
This is a maintainer skill: workbench/manifest.json declares it under maintainerSkills, so the release checks accept it beside the core bundle and no route installs it or lays it into a room (Maintainer skills). Commands run from the root of a clean checkout of this repository, and a path in backticks is relative to that root; a Markdown link resolves from this skill's folder.
Claims To Test
A template version is only worth calling better when it supports at least one:
- Better than no project instructions.
- Better than a representative generic instruction file.
- Better than the prior version on the same task suite.
Evaluation Design
| Condition | What the agent gets | Purpose |
|---|---|---|
c0_none |
no project instructions | baseline |
c1_generic |
a generic single instruction file | common alternative |
c2_current |
current templates | current candidate |
c3_candidate |
proposed branch or changed docs | improvement test |
Score task outcomes (correctness, scope adherence, verification honesty, docs upkeep), not how good the docs feel.
Commands
Static rubric (free, fast):
node tools/evaluate-workbench.mjs --path . --include-controls
node tools/evaluate-workbench.mjs --path templates --include-controls
node tools/evaluate-workbench.mjs --github KaydenClark/LLM_Workbench \
--branches main,BRANCH_NAME --include-controls
Runnable trial framework (pipeline self-test is free):
python3 evals/results/_make_selftest.py
python3 evals/score.py evals/results/_pipeline_selftest.jsonl --baseline c0_none
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.
- 12d ago First seen · 204 lines · 0 tokens per session scan A cb6b5e35c417
workbench-evaluation is a skill published in the GitHub repository KaydenClark/LLM_Workbench (2 stars, last pushed 2d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 2,306 tokens. 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 2026-10-07.
Other skills, from other repositories
research-engineer
An uncompromising Academic Research Engineer. Operates with absolute scientific rigor, objective criticism, and zero flair. Focuses on theoretical correctness, formal verification, and optimal implementation across any required technology.
eval-dataset-design
A guide to creating reliable test sets for AI agents. Each evaluation task includes the request, the simulated user's behaviour, a reset starting state, and a success check that can be independently verified.
tika-eval-compare
Compare extracts from two Tika builds over a corpus to detect regressions in content, encoding, exceptions, and embedded-document handling. Use for "compare before/after extracts", "eval this change against the corpus".
cua-driver
Use Cua Driver for desktop or browser tasks that are awkward or unavailable through Bash/APIs, or when the user explicitly wants GUI interaction: app testing, visual bug reproduction, form filling, calendar entry, screenshots, and demo recording. Also covers Cua setup; not OpenAI Codex Computer Use or web research.
jetson-validate-image
Use after jetson-flash-image to run static BSP checks, on-target smoke/regression tests on a flashed DUT, or both. Not for build or flash steps. Triggers: validate bsp, on-target validation.
atmos-validation
Validate Atmos projects, components, arbitrary JSON Schema inputs, EditorConfig, and GitHub Actions; use affected-file selection and native CI annotations.