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 TimothyHan/qa-buddy-skills --skill improvegit clone --depth 1 https://github.com/TimothyHan/qa-buddy-skillsWrote 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/timothyhan/qa-buddy-skills/improve)<a href="https://agentmods.dev/skills/timothyhan/qa-buddy-skills/improve"><img src="https://agentmods.dev/badge/skills/timothyhan/qa-buddy-skills/improve/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/timothyhan/qa-buddy-skills/improve"><img src="https://agentmods.dev/badge/skills/timothyhan/qa-buddy-skills/improve.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00169 | $0.03990 |
| Opus 5.5 | $0.00068 | $0.01596 |
| Sonnet 5.5 | $0.00034 | $0.00798 |
| Haiku 4.5 | $0.00017 | $0.00399 |
Grade A, and why
improve 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 1mo 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 — 298 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/qa-improve: Skill Self-Improvement
You are a meta-skill with two modes:
- Fix mode (default, Phases 1–6): an SDT reports a skill produced incorrect, misleading, or incomplete output — analyze root cause, propose targeted fixes, apply, verify.
- Distill mode ("distill learnings", a skill flagged a falsified entry, or the
learnings file exceeds ~30 active entries): maintain the project learnings layer —
see the Distill Mode section at the end.
--dry-run(or "propose only") writes the sweep as a proposal file and edits nothing.
Constraints
- Never guess the failure. Ask the SDT what happened and what they expected. Don't infer from vague descriptions.
- Read the skill before proposing changes. Every fix must reference the specific line, phase, or section it changes.
- Follow CONTRIBUTING.md. Read it before making changes. Respect the 300-line skill budget, Sonnet context limits, and structural conventions.
- One root cause at a time. If the SDT reports multiple issues, address them sequentially. Don't batch unrelated fixes.
- Preserve workflow-specific knowledge. Never remove phase logic, KB paths, severity scales, or SDT approval gates as part of a trim.
- Test the fix conceptually. Before applying, explain how the fix prevents the original failure from recurring.
- Bump the version. Every skill change gets a patch or minor version bump.
Phase 1: Understand the Failure
Ask the SDT these questions (skip any the SDT already answered):
- Which skill? — Which skill produced the bad output? (e.g.,
/qa-test-plan,/qa-qa) If unclear, check the recent session for skill invocations. - What happened? — What did the skill output? Be specific: quote the bad text, show the wrong table, describe the missing section.
- What was expected? — What should the output have been? What would correct behavior look like?
- How was it caught? — Did the SDT notice during review? Did another skill fail because of it? Did a team member flag it?
- Is this repeatable? — Would the same inputs produce the same bad output every time, or was it a one-off?
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.
- 1mo ago Changed e6191492cfe5
- 1mo ago Changed · +5 lines ff651050518d
- 1mo ago First seen · 293 lines · 169 tokens per session scan A e39dc3b09d2b
improve is a skill published in the GitHub repository TimothyHan/qa-buddy-skills (9 stars, last pushed 20d ago), licensed Apache-2.0. It adds 169 tokens to every session and 3,990 once invoked, about $0.0007 per session on Opus 5.5. 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 2026-08-31.
Other skills, from other repositories
agent-memory-discipline
Rules for when an agent should recall from long-term memory before acting and when it should save decisions, corrections and failures afterwards. Works with any memory backend.
agent-state-bar
A design guide for an agent status bar: a short, automatically updated summary of progress, remaining tasks, tool counts, errors, and system state placed near the latest model input.
kv-cache-design
A guide to arranging an AI agent's conversation so previously processed text can be reused. A KV cache stores intermediate results for earlier tokens, while a prompt cache reuses a stable beginning across requests.
reward-design
A guide to defining the scores that an AI agent receives during trial-and-error training. It covers checks based on rules, human preferences, or another model, and distinguishes rewarding the final result from rewarding the steps taken.
agent-evolution
A guide to making an AI agent improve from its past runs. It covers how to judge results, extract reusable lessons, and decide whether to store them as knowledge, instructions, code, or model training.
knowledge-org
A guide to organising large bodies of knowledge so an AI system can navigate relationships, summaries, and layers instead of searching only flat text chunks. It covers tree indexes, knowledge graphs, file-like structures, incremental updates, and long-term memory layouts.