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/akaszubski/autonomous-dev/skill-evalgit clone --depth 1 https://github.com/akaszubski/autonomous-devWhat 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 | $0.00013 | $0.00539 |
| Opus 5 | $0.00006 | $0.00269 |
| Sonnet 5 | $0.00003 | $0.00108 |
| Haiku 4.5 | $0.00001 | $0.00054 |
Grade A, and why
skill-eval 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 2d 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
No licence file
A repository with no LICENSE is all rights reserved by default, so the body is not copied here. The metadata, the measurements and the link are.
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.
- 2d ago First seen · 59 lines · 13 tokens per session scan A 3f543ef9670f
skill-eval is a command published in the GitHub repository akaszubski/autonomous-dev (33 stars, last pushed 2d ago), with no licence file. It adds 13 tokens to every session and 539 once invoked, about $0.0001 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 commands, from other repositories
VibeGuard: Learn
Dual-mode learning: (A) Extract guard rules from errors (B) Extract reusable skills from discoveries. Automatic mode routing.
VibeGuard: Skill Validate
Gate proposed skills with with/without repair and regression evidence.
trace
Render the reasoning trace for a task — filereads, outputclaims, toolcalls, decisions. Surfaces hallucinations (claims with grounded: false).
loop
The loop pack is dxkit's deterministic preflight/postflight layer for autonomous coding loops — a Claude Code session that keeps working until it decides to stop. It stops a loop from declaring "done" while it has introduced net-new findings, by re-running the guardrail on every Stop and feeding any net-new findings…
polygen
Run polygen — draft a contract from a feature description, author a verifiable SAM v2 strict-profile module against it, self-repair against reachable invariant violations, and synthesize a demo/regression trace corpus.
workflow
Build a feature as a workflow, not an agentic loop — design stages/signals/budgets with the user, front-load contract + invariants, author with polygen, declare effects, prove the anti-loop rules with check-effects, and stand it up under polyrun.