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 commands/rhyanvargas/agentic-development-starter-kit/run-skill-evalsgit clone --depth 1 https://github.com/rhyanvargas/agentic-development-starter-kitWrote 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/commands/rhyanvargas/agentic-development-starter-kit/run-skill-evals)<a href="https://agentmods.dev/commands/rhyanvargas/agentic-development-starter-kit/run-skill-evals"><img src="https://agentmods.dev/badge/commands/rhyanvargas/agentic-development-starter-kit/run-skill-evals.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 | $0.00000 | $0.00637 |
| Opus 5 | $0.00000 | $0.00318 |
| Sonnet 5 | $0.00000 | $0.00127 |
| Haiku 4.5 | $0.00000 | $0.00064 |
Grade A, and why
run-skill-evals 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 4d 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 — 38 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/run-skill-evals
Run with-skill vs without-skill output evals for a skill (Tier 2 style). Soft quality signal — not a merge gate.
Skill / playbook
Follow skills/skill-optimizer → references/eval-loop.md (or .agents/skills/skill-optimizer in adopter apps). Full runbook: docs/evaluating-skills.md.
Usage
/run-skill-evals
/run-skill-evals skills/spec-driven-workflow
/run-skill-evals .agents/skills/my-company-skill
When to use
- You changed a skill’s behavior or description, or switched models
- You are shipping a company skill and need with/without evidence before commit
Do not re-benchmark every PR. For stock first-party skills, trust published numbers in docs/evals/SCORECARD.md unless the above applies.
Behavior
- Locate the target skill (ask if unclear). Prefer kit
skills/<name>or adopter.agents/skills/<name>. - Confirm
evals/evals.jsonexists (and ideallyevals/trigger/eval_queries.json). If missing, stop and point to/optimize-skillfirst. - Package cases (pick one):
- This kit repo: run
./scripts/run-skill-evals-soft.sh --skill <name>→ work under.adsk-tier2-out/<name>/. - Adopter app (no kit scripts): read prompts from
.agents/skills/<name>/evals/evals.jsondirectly; create a gitignored workspace (e.g..adsk-tier2-out/<name>/) withwith_skill/andwithout_skill/arms per case.
- This kit repo: run
- For each eval case, in a clean context:
- with_skill — skill available; paste the prompt; save output
- without_skill — same prompt without the skill (or prior version); save output
- Grade each assertion PASS/FAIL with evidence into
grading.json(or equivalent). Prefer mechanical checks when possible. - Summarize pass-rate Δ (and token/time Δ if available). For kit first-party skills, offer a SCORECARD paste row; for company skills, report in-chat / PR notes (do not invent kit SCORECARD edits unless asked).
- Required close-out: load
skill-optimizer→references/eval-loop.md→ After grading — recommended next actions. Append a Recommended next actions section (Do now / Do next / Skip) mapped from the decision table — fixwith_skillFAILs first, then SCORECARD/PR notes, then assertion tighten. Offer to execute the top action when it is a small obvious edit. Do not end on scores alone. - Reminder: Tier 1 (
./scripts/check-skills-ci.shin the kit) only checks harness shape — it does not replace this loop.
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.
- 4d ago First seen · 38 lines · 0 tokens per session scan A addd90c9433c
run-skill-evals is a command published in the GitHub repository rhyanvargas/agentic-development-starter-kit (2 stars, last pushed 11d ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 637 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 2026-08-31.
Other commands, from other repositories
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
specify
Create or update the feature specification from a natural language feature description.
converge
Assess the current codebase against the feature's spec, plan, and tasks, then append any remaining unbuilt work as new tasks to tasks.md so implement can complete it.
implement
Execute the implementation plan by processing and executing all tasks defined in tasks.md.
analyze
Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md after task generation.