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/luiseiman/dotforge/capture-practicenpx skills add luiseiman/dotforge --skill capture-practicegit clone --depth 1 https://github.com/luiseiman/dotforgeWhat 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 | $0.00024 | $0.00851 |
| Opus 5 | $0.00012 | $0.00426 |
| Sonnet 5 | $0.00005 | $0.00170 |
| Haiku 4.5 | $0.00002 | $0.00085 |
Grade A, and why
capture-practice 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.
How it starts
The opening of the file, as written. The whole thing — 119 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Capture Practice
Register an insight, pattern, or lesson learned in the dotforge inbox.
Input routing
If $ARGUMENTS is empty → auto-detect mode (Step 0). If $ARGUMENTS has content → skip to Step 1.
Step 0: Auto-detect (only when $ARGUMENTS is empty)
Analyze the recent conversation to identify the most generalizable insight from this session.
Detection signals (in priority order)
- Workaround discovered — the obvious approach failed and an alternative was used
- Multi-attempt bug — a fix required more than one attempt or root cause was non-obvious
- Architectural decision with trade-offs — two+ alternatives were considered, one chosen with reasoning
- Non-obvious tool/API/flag behavior — a parameter, edge case, or behavior was surprising
- Missing rule — the session revealed a gap in
.claude/rules/orCLAUDE.mdthat would have prevented the problem
Extraction rules
- Extract the single most generalizable insight — not session notes, not a summary of everything done
- Formulate it as a reusable principle: "When X, do Y because Z" or "Never do X — use Y instead"
- Keep it to 1-2 sentences max
- Ignore: trivial tasks, first-attempt successes, routine edits
If no signal is present
Respond: "No generalizable insight detected in this session. If you have something specific in mind, run /cap \"description\"."
Stop — do not create a file.
Propose and confirm
Show the proposed insight and ask for confirmation before proceeding:
Proposed practice:
"{{one-line insight}}"
Tags: {{inferred tags}}
Project: {{current project name}}
Save this? [Y/n/edit]
- If Y or user confirms → continue to Step 1 with the proposed text as $ARGUMENTS
- If n → stop, no file created
- If edit or user rewrites → use the rewritten text as $ARGUMENTS, continue to Step 1
Step 1: Parse the insight
From $ARGUMENTS (provided or confirmed from Step 0), extract:
- What: the practice or pattern
- Why: evidence or context (current project, error that motivated it)
- Impact: which dotforge files could change
- Tags: categorization (hooks, rules, prompting, security, stack-specific, etc.)
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 · 119 lines · 24 tokens per session scan A f7ec41100fad
capture-practice is a skill published in the GitHub repository luiseiman/dotforge (8 stars, last pushed 2mo ago), licensed MIT. It adds 24 tokens to every session and 851 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 2026-08-31.
Other skills, from other repositories
lesson-learned
Extract learnings and save as memory files (file-based memory system). Use when: "lesson learned", "what did we learn", "learning", "remember".
qc-helper
Answer any question about Qwen Code usage, features, configuration, and troubleshooting by referencing the official user documentation. Also helps users view or modify their settings.json. Invoke with /qc-helper followed by a question, e.g. /qc-helper how do I configure MCP servers? or /qc-helper change approval mode…
course-storytelling
Use when lesson or course content is correct but forgettable and a concept has to LAND: profiles the learner, breaks the blocking false belief, then rebuilds it as epiphany story → named model → grounded analogy → proof → so-what. NOT outcomes, assessment or module order (that is course-builder), NOT slide visuals…
eli5
Explain a topic like I'm a 5 year old — restate my last output, or a named topic, in plain words without dropping a single fact. Use when the user types /eli5 [topic], or says an answer was too technical, too long, or unclear about what he now has to do.
anthropic-skilljar-extractor
Extracts any Anthropic Academy (Skilljar) course into a structured markdown folder. Triggers when the user provides an anthropic.skilljar.com URL and asks to extract, research, download, or convert a course to markdown. Produces one markdown file per lesson organized into section folders, with course diagrams…
backend-interview
模拟后端技术面试，基于简历生成针对性问题.