메뉴
Add words/terms you're learning, enriched (translation, alternative translations, examples, synonyms, definition) into your native/learning languages. Comma-separated for several at once. Usage: /loot <word>[, <word2>, ...]
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Dev: check a target (spec / plan / working diff / code) for architectural drift from the rulebook docs/ARCHITECTURE.md — report qualitative deviations citing rule ids, and triage new ones into docs/Debt.md. Usage: /drift <target>
Push enriched vocab to Anki Desktop as flashcards and pull review progress back. Usage: /anki-sync
Review your buffered writing into per-category frequency docs (grammar / rephrasings / idioms / verbs). Usage: /debrief
Use whenever you read, query, or mutate shadowling's sqlite database (shadowling.db) — any SELECT/INSERT/UPDATE/DELETE, ad-hoc SQL, dropping/clearing/resetting rows, inspecting data, or checking a migration's result. All DB access goes through a skill entrypoint, a repository method (`models/*`), or `sql.py` — never raw sqlite3. Schema/data-layer changes are NOT made here — they go through an appended migration in appdb.py; this skill only documents that rule and verifies the outcome.
Explain an expression in the language you're learning that you can't read literally — get a verdict (memorize vs learnable rule) + how to read it, saved to the decode dataset. Usage: /aha <phrase> [+ your hunch]
Remove words/terms from your vocab list. Comma-separated for several at once. Usage: /drop <word>[, <word2>, ...]
| name | loot |
| description | Add words/terms you're learning, enriched (translation, alternative translations, examples, synonyms, definition) into your native/learning languages. Comma-separated for several at once. Usage: /loot <word>[, <word2>, ...] |
| agent | claude |
| allowed-tools | Bash(python3 */loot.py*) |
You enrich the given terms and add them to the vocab store. Add everything you are given — do NOT ask, do NOT block on typos. This skill runs in the MAIN context so you can read the conversation: that is where each word's real usage context lives.
The driver is ${CLAUDE_PLUGIN_ROOT}/loot.py. Invoke it directly so the command
starts with python3. It gates on config itself — if config is missing it prints a
setup notice and exits non-zero; just relay that (step 4).
Terms (comma-separated): $ARGUMENTS
Steps:
Split $ARGUMENTS on commas, trim each, drop empties. Keep multi-word phrases
whole (machine learning is one term). Lowercase each term.
For EACH term separately, decide its micro-context <ctx> — a short real
snippet from THIS conversation that illustrates how that term was actually used
(enough to show its meaning in this context), authored fresh per term. Include
<ctx> ONLY when the conversation genuinely holds such a usage for that specific
term; a sentence that merely contains the token, or no real encounter at all,
does NOT qualify. When in doubt, or for an ad-hoc add, OMIT the <ctx> tag
entirely — the driver then generates a generic example.
ONE call. Pipe the words as a tagged <items> block on stdin via a quoted
heredoc. One <row> per term: <word> is the term; add a <ctx> child ONLY
when you have a genuine usage snippet for it, otherwise omit the tag. The body is
well-formed XML — escape & as &, < as <, > as > inside
values; the quoted <<'SL_IN' handles quotes/$/backticks. The body and the
closing SL_IN MUST start at column 0:
python3 "${CLAUDE_PLUGIN_ROOT}/loot.py" <<'SL_IN'
<items>
<row><word>throughput</word><ctx>We boosted throughput under load.</ctx></row>
<row><word>idempotent</word></row>
</items>
SL_IN
Relay the driver's summary line verbatim (N/M enriched; re-run /loot to retry …
if any stayed pending). If it printed a config notice instead, point the user at
/shadowling:setup.
Do NOT gloss anything in your reply.