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 skills/itaywol/adeptability/adept-self-improvenpx skills add itaywol/adeptability --skill adept-self-improvegit clone --depth 1 https://github.com/itaywol/adeptabilityWrote 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/itaywol/adeptability/adept-self-improve)<a href="https://agentmods.dev/skills/itaywol/adeptability/adept-self-improve"><img src="https://agentmods.dev/badge/skills/itaywol/adeptability/adept-self-improve.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 5.1 | $0.00048 | $0.00588 |
| Opus 5 | $0.00024 | $0.00294 |
| Sonnet 5 | $0.00010 | $0.00118 |
| Haiku 4.5 | $0.00005 | $0.00059 |
Grade A, and why
adept-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 6d 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 — 57 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Self-improve: turn lessons into skills
adept skills are portable memory. A lesson written as a skill is loaded by every harness on every future session — not lost when this context ends. When you learn something durable, capture it instead of forgetting it.
When to capture
Capture when the lesson is durable and reusable, not one-off:
- The user corrects you ("we always use X, not Y", "don't touch Z").
- A fix or pattern generalizes beyond the file you're in.
- You discover a project convention the code doesn't make obvious (a build gate, a naming rule, a required pre-PR step).
- You catch yourself thinking "I'll remember that for next time." — you won't; the skill will.
Don't capture: secrets, one-shot facts, anything already obvious from the code or
existing skills. Improve an existing skill before adding a near-duplicate (adept skill list).
The loop
adept skill list # does a skill already cover this? edit it if so
adept skill add <kebab-id> --edit # else scaffold a new one
# write a triggering description + a tight, harness-neutral body
adept sync # render to every enabled harness
adept status && adept diff # confirm it landed clean
Then tell the user, in one line, what you captured and where — they own the memory, so let them veto it. Commit the new skill with their normal review.
What makes a captured lesson good
- Description = trigger. Front-load when it applies so a future agent loads it at the right moment. See [[authoring-adept-skills]] for crafting the description and activation.
- Atomic. One lesson per skill. If it needs "and also…", it's two skills.
- Reason, not just rule. "Use sentinel errors so callers can
errors.Is" beats "use sentinel errors" — the why survives refactors the rule wouldn't.
Confirm before you persist
Capturing is an outward, durable act: it changes how every future session behaves. Surface what you're about to write and let the user confirm — don't silently rewrite the project's shared memory. See [[using-adept]] for the surrounding CLI.
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.
- 6d ago First seen · 57 lines · 48 tokens per session scan A 9e5762f3a54c
adept-self-improve is a skill published in the GitHub repository itaywol/adeptability (9 stars, last pushed 1mo ago), licensed MIT. It adds 48 tokens to every session and 588 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 2026-08-31.
Other skills, from other repositories
api-baas-appwrite
Appwrite backend-as-a-service — Auth, TablesDB, Storage, Functions, Realtime, permissions model, typed client.
api-database-mongodb
Native MongoDB driver (the mongodb npm package) - MongoClient lifecycle, typed collections, CRUD result shapes, cursors, aggregation pipelines, index design, transactions.
web-meta-framework-sveltekit
SvelteKit full-stack framework - file-based routing, load functions, form actions, server hooks, SSR/SSG, API routes, streaming, progressive enhancement.
ai-infrastructure-ollama
Local LLM inference with the Ollama JavaScript client -- chat, streaming, tool calling, vision, embeddings, structured output, model management, and OpenAI-compatible endpoint.
ai-orchestration-langchain
LangChain.js patterns for building LLM applications — chat models, LCEL chains, prompt templates, structured output, agents, tools, RAG, streaming, and LangSmith tracing.
ai-orchestration-vercel-ai-sdk
Vercel AI SDK patterns - providers, text generation, streaming, structured output, tool calling, chat UI hooks, embeddings, and RAG.