From rocket
Wipe the project's rocket-managed contextualization files (lexicon, logs, locks) under .roc/rocket/. Use this skill whenever the user invokes "/rocket:context-clear", says "clear context", "reset the lexicon", "wipe the lexicon", "clean context files", "nettoie le contexte", "supprime les fichiers de contextualisation", "réinitialise le lexique", "vide le lexique", "remets le contexte à zéro", or any similar request to discard the rocket contextualization state for the current project.
How this skill is triggered — by the user, by Claude, or both
Slash command
/rocket:context-clearThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Wipe the rocket contextualization files in the current project: the lexicon, the wrapper logs, leftover lock dirs, and any leftover atomic-write temp files. Two-step UX: first invocation lists, second invocation with `force` deletes.
Wipe the rocket contextualization files in the current project: the lexicon, the wrapper logs, leftover lock dirs, and any leftover atomic-write temp files. Two-step UX: first invocation lists, second invocation with force deletes.
Strictly the current project, under .roc/rocket/:
.roc/rocket/lexicon.md — the project lexicon..roc/rocket/lexicon.md.tmp — leftover atomic-write temp file (if any)..roc/rocket/lexicon-update.log, .log.1, .log.2, .log.3 — wrapper logs (rotated)..roc/rocket/lexicon.md.lock.d/ — atomic lock directory (transient)..roc/rocket/ itself — removed only if empty after cleanup..roc/ itself — removed only if empty (other plugins may still use it).This skill does not touch:
CLAUDE.md — the ## Project semantic context block stays. To remove it, edit CLAUDE.md manually.~/.roc/... — user-global state (other plugins, voice profiles, mail state). Out of scope..roc/<other-plugin>/.The argument from the user is $ARGUMENTS (whitespace-trimmed, lowered).
ls -la .roc/rocket/ 2>/dev/null as a single Bash call to enumerate what is there..roc/rocket/ does not exist or is empty, print exactly: No rocket contextualization files in this project. and stop.Run /rocket:context-clear force to actually delete these files. Re-run /rocket:setup afterwards if you want to re-bootstrap the lexicon.
force — delete modeIssue every Bash call as a single atomic command. Do not chain with &&, ||, ;. Each step below is one separate Bash invocation.
rm -f .roc/rocket/lexicon.mdrm -f .roc/rocket/lexicon.md.tmprm -f .roc/rocket/lexicon-update.logrm -f .roc/rocket/lexicon-update.log.1rm -f .roc/rocket/lexicon-update.log.2rm -f .roc/rocket/lexicon-update.log.3rmdir .roc/rocket/lexicon.md.lock.d/ (Bash exit code may be non-zero if missing — fine).rmdir .roc/rocket/ (Bash exit code may be non-zero if other rocket files remain — fine).rmdir .roc/ (Bash exit code may be non-zero if other plugins still occupy it — fine).Cleared rocket context files for this project. Run /rocket:setup to re-bootstrap if needed.Print exactly: Unknown argument "<arg>". Use /rocket:context-clear (list mode) or /rocket:context-clear force (delete). and stop.
rm -rf .roc/ indiscriminately. Other plugins may live there. Step 7-9 above use plain rmdir, which fails (and that's fine) when the directory still has content.CLAUDE.md.~/.roc/... (user-global state).User: /rocket:context-clear
total 32
-rw-r--r-- 1 user staff 8217 May 4 13:00 lexicon.md
-rw-r--r-- 1 user staff 1190 May 4 13:30 lexicon-update.log
Run /rocket:context-clear force to actually delete these files. Re-run /rocket:setup afterwards if you want to re-bootstrap the lexicon.
User: /rocket:context-clear force
Cleared rocket context files for this project. Run /rocket:setup to re-bootstrap if needed.
/rocket:context-clear without the explicit force confirmation step. Always two-step.CLAUDE.md to remove the ## Project semantic context section. Out of scope..roc/ with rm -rf. Other plugins may occupy it.~/.roc/rocket/ (user-global). This skill is project-scoped.npx claudepluginhub digivorefr/roc --plugin rocketGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.