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/ShaftHQ/SHAFT_ENGINEnpx agentmods add skills/shafthq/shaft_engine/self-improveWrote 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/shafthq/shaft_engine/self-improve)<a href="https://agentmods.dev/skills/shafthq/shaft_engine/self-improve"><img src="https://agentmods.dev/badge/skills/shafthq/shaft_engine/self-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/shafthq/shaft_engine/self-improve"><img src="https://agentmods.dev/badge/skills/shafthq/shaft_engine/self-improve.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 | $0.00033 | $0.01330 |
| Opus 5.5 | $0.00013 | $0.00532 |
| Sonnet 5.5 | $0.00007 | $0.00266 |
| Haiku 4.5 | $0.00003 | $0.00133 |
Grade A, and why
self-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 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 — 122 lines — stays where its author put it; the contents beside it link to each section on GitHub.
self-improve — ChaosEngine learning & adapting
Lean CE-native skill informed by Task Observer methodology (Eoghan Henn / rebelytics, CC BY 4.0 — see LICENSE and UPSTREAM.md). Not a blind clone.
When
- Primary: root-owned Learning Session after every final delivery (owner approval, publish done, final PR merged), automatic: lessons diff against the skills, one spec issue, ONE PR (
skip-release-notes, auto-merge MERGE) babysat to merged. - Secondary: explicit operator request mid-session.
- Not: every casual turn, and never the Learning Session PR's own merge.
- A lesson already shipped in the delivery pull request is not a new issue. Finalize with nothing durable.
If assess refuses novel_success under the one-incident rule, still
finalize: file or update the durable lesson tickets, then complete finalize
without inventing a second incident for the same success signal.
Dual track
- Harness — skills, hooks, MemPalace, Graphify, installer/doctor. GitHub issue only. Never a local queue and never chat.
- Product — enhancements for the product under development (queued issues).
Details: references/observation-taxonomy.md. Product track: references/product-track.md (product issues/evals + ChaosGauge link; CLI/doctor/silent-verify over essay tickets). Activation: references/activation.md. Adopt/reject research: references/research-adopt-reject.md. Roadmap: ../../references/self-improve-master-plan.md. Research/explore isolation (harness): ../../references/context-firewall.md.
How (wraps learning.py)
- Classify each finding (harness vs product; category allow-list).
- Write minimal fields only:
category,title,lesson,proposedChange,benefit,estimatedTokens. - File a harness lesson as a GitHub issue. Queue only a product lesson
through
learning.py --track product. - Never auto-install skill patches; stage proposals for human/CI review.
- "Nothing durable" is a valid outcome.
- A lessons pull request lands only when it states the concrete outcome the
lesson would have changed: time, a defect, or a failure avoided. The body
has a
## Outcome changedsection with that statement. A title starting withLessonorLessonsis a lessons PR. The PR body gate rejects one that lacks the section. A candidate that fails the bar is one append-only line in.chaos-engine-state/lesson-notes.log(record_lesson_note), not a pull request. Zero-trigger rule retirement is rule-retirement. - Refine this protocol when traces show skipped Learning Sessions, local writers reviewing their own diffs, or queue-only lessons that never became issues. Self-development has no cap.
What ships with it
6 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.
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 Changed · +8 lines 42a5954de2c1
- 15d ago Changed · +2 lines 2258d90e9be0
- 18d ago Changed · +1 lines 34a9549dff93
- 23d ago Changed · +4 lines d7763e2d8901
- 25d ago Changed · -6 lines · +3 tokens per session d4b698f4bc93
- 29d ago Changed b27470042b70
- 1mo ago Changed · +20 lines · -2 tokens per session 6f2c8c787a56
- 1mo ago Changed · +3 lines 29435fd29662
- 1mo ago Changed 84e11951a7f6
- 1mo ago First seen · 90 lines · 32 tokens per session scan A 0ae43b1ff610
self-improve is a skill published in the GitHub repository ShaftHQ/SHAFT_ENGINE (417 stars, last pushed today), licensed MIT. It adds 33 tokens to every session and 1,330 once invoked, about $0.0001 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 2026-09-07.
Other skills, from other repositories
Framework Architect
Skill thiết kế và scaffold automation framework hoàn chỉnh cho Playwright, Selenium, và Appium — bao gồm project structure, base classes, config management, reporting, và CI/CD integration.
flowproof-config
Configure flowproof's SAP GUI, Fiori, and AI authoring credentials by walking the user through flowproof config sap / flowproof config fiori / flowproof config ai. Use when the user wants to set up, change, or check their SAP/Fiori login or model authoring key, or when a flow run fails because SAPUSER, FIORIPASSWORD…
Smart Locator Agent
Skill sinh locator ổn định và dễ bảo trì cho UI automation, hỗ trợ Playwright, Selenium và Appium.
QA Automation Engineer
Skill hỗ trợ agent thực hiện các tác vụ QA automation testing bao gồm generate test cases, automation scripts, API tests, locators, phân tích flaky tests, và tạo test data.
Locator Healer Agent
Skill tự động phát hiện và sửa chữa locators bị hỏng khi automation tests fail do thay đổi DOM.
tool-assistant
Tool-calling assistant with backend functions for time, city time, weather, and temperature conversion, plus cost metering and approval workflows. Use when a question needs real-time data via tool calls.