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 jikig-ai/soleur --skill eval-harnessgit clone --depth 1 https://github.com/jikig-ai/soleurWrote 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/jikig-ai/soleur/eval-harness)<a href="https://agentmods.dev/skills/jikig-ai/soleur/eval-harness"><img src="https://agentmods.dev/badge/skills/jikig-ai/soleur/eval-harness/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/jikig-ai/soleur/eval-harness"><img src="https://agentmods.dev/badge/skills/jikig-ai/soleur/eval-harness.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 | $0.00036 | $0.02573 |
| Opus 5 | $0.00018 | $0.01287 |
| Sonnet 5 | $0.00007 | $0.00515 |
| Haiku 4.5 | $0.00004 | $0.00257 |
Grade A, and why
eval-harness 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 5d 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.
The source is not reproduced here
A licence we could not identify
The repository carries a LICENSE file, but it is custom or dual enough that GitHub cannot name it and neither can this catalogue. Unknown terms are not permission, so the body is not copied here. Read the licence at the source and decide for yourself.
What ships with it
44 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.
- enums/go-routes.json 119 B
- enums/incident-threshold.json 62 B
- enums/lane.json 56 B
- enums/tool-selection.json 300 B
- enums/triage-levels.json 27 B
- gated-skills.json 1.5 KB
- models.generated.json 131 B
- promptfooconfig.go-routing.yaml 1.8 KB
- promptfooconfig.incident-threshold.yaml 1.5 KB
- promptfooconfig.lane-inference.yaml 1.6 KB
- promptfooconfig.ticket-triage.yaml 1.6 KB
- promptfooconfig.tool-selection.yaml 1.7 KB
- prompts/go-baseline.txt 294 B
- prompts/go-skill.txt 3.1 KB
- prompts/incident-threshold-baseline.txt 280 B
- prompts/incident-threshold-skill.txt 721 B
- prompts/lane-inference-baseline.txt 282 B
- prompts/lane-inference-skill.txt 1.4 KB
- prompts/tool-selection-baseline.txt 1.0 KB
- prompts/tool-selection-skill.txt 1.2 KB
- prompts/triage-baseline.txt 215 B
- prompts/triage-skill.txt 440 B
- README.md 10 KB
- scripts/eval-gate.cjs 14 KB runs code
- scripts/extract-block.cjs 3.0 KB runs code
- scripts/gate-classification.cjs 1.2 KB runs code
- scripts/gen-models.sh 1.5 KB runs code
- scripts/gen-skill-prompt.cjs 6.0 KB runs code
- scripts/measure-classification.cjs 1.3 KB runs code
- scripts/parse-label.cjs 4.0 KB runs code
- scripts/verdict.cjs 5.8 KB runs code
- tasks/go-routing.jsonl 1.1 KB
- tasks/incident-threshold.jsonl 1020 B
- tasks/lane-inference.jsonl 900 B
- tasks/ticket-triage.jsonl 1.4 KB
- tasks/tool-selection.jsonl 3.5 KB
- test/eval-gate.test.sh 6.0 KB runs code
- test/extract-block.test.sh 3.8 KB runs code
- test/gate-classification.test.sh 3.5 KB runs code
- test/gen-models.test.sh 2.9 KB runs code
- test/measure-classification.test.sh 3.6 KB runs code
- test/parse-label.test.sh 6.5 KB runs code
- test/registry-completeness.test.sh 5.7 KB runs code
- test/verdict.test.sh 5.9 KB runs code
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.
- 5d ago First seen · 146 lines · 36 tokens per session scan A ecdeb3a00967
eval-harness is a skill published in the GitHub repository jikig-ai/soleur (15 stars, last pushed today), with no licence file. It adds 36 tokens to every session and 2,573 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 2026-09-04.
Other skills, from other repositories
skill-coverage-audit
Trace codepaths in diffs, map against tests, auto-generate missing coverage — use before shipping PRs.
skill-verification-gate
Use when about to declare work complete, fixed, passing, or done.
skill-tdd
Build a behavior change with observed red, minimal green, and measured test consolidation.
skill-verify
Use when a nontrivial change needs end-to-end verification before committing or shipping.
eval
This skill should be used to make eval the unit of progress: systematize how a feature or agent is judged. Trigger with "how do we eval this", "build an eval harness", "why is the agent failing", "error analysis", "are we regressing", or before flipping an eval-gated dark launch. Builds evals from REAL failures (not…
writing-skills
Use when creating a new forge-skill, editing an existing skill, when planning to contribute to forge-skills, when a session has produced a one-off prompt that should be promoted to a skill, or when verifying a skill works before deployment.