tokensaver

Plugin tokensaver from marketplace tokensaver-marketplace in Brucee/tokensaver. Commands read from the repo's manifest — 16 days ago.

Skills (5)

• tokensaver-review

  Local multi-lens code review: collects the full change (committed vs base, staged, unstaged, untracked) and fans it out to parallel single-lens review subagents, then mechanically verifies citations and synthesizes one ranked report. Use when the user asks for a code review, a review before committi

• tokensaver-savings

  How tokenSaver measures savings — the equivalent-calls model, pricing table, persisted state files, and how to read them. Use when the user asks about the savings numbers, the status line figures, or the state files.

• tokensaver-settings

  Change tokenSaver settings — enable/disable the plugin features, the status line, hook behavior. Use when the user asks to turn something on/off, change plugin behavior, or asks what settings exist.

• tokensaver-status

  Check tokenSaver's health — version, data dir, settings, lifetime savings, and the doctor self-check for broken installs. Use when the user asks if the plugin is working, reports that hooks/tools/savings stopped, or wants diagnostics.

• tokensaver

  What the tokenSaver plugin does, its tools, commands and configuration. Use when the user asks about tokenSaver itself, its features, or how to control it.

Slash commands (1)

• /tokensaver:update

  Update tokenSaver to the latest marketplace release (restart applies it)

Subagents (4)

• code

  tokenSaver enhanced coding agent with smart search, batch editing, and SQL introspection. Use as the default main thread agent.

• explore

  Fast read-only agent for file searches, symbol lookups, and codebase questions like "where is X defined?", "where is X called?", or "how does X flow through the system?". Prefer for exploration requiring several Search calls. Uses haiku for focused lookups.

• recall

  Runs the recall CLI over the user's past Claude Code sessions and returns the one actionable item — a command, decision, or fix — distilled from a prior conversation, not raw transcripts.

• review

  Single-lens code reviewer. The orchestrator (tokensaver-review skill) launches several of these in parallel, each with one review lens. Cheap model (haiku) — a narrow lens does not need a frontier model.

MCP servers (1)

• code

  runs: node --no-warnings=ExperimentalWarning -e var a=process.argv[1]||'';require(((a.indexOf('${')<0&&a)||process.cwd())+'/scripts/ensure-deps.cjs') ${CLAUDE_PLUGIN_ROOT}

Hooks (9)

• SessionStart

  node --no-warnings=ExperimentalWarning "${CLAUDE_PLUGIN_ROOT}/scripts/session-hook.cjs"

• UserPromptSubmit

  node --no-warnings=ExperimentalWarning "${CLAUDE_PLUGIN_ROOT}/scripts/session-hook.cjs"

• PreToolUse

  node --no-warnings=ExperimentalWarning "${CLAUDE_PLUGIN_ROOT}/scripts/session-hook.cjs"

• PostToolUse

  node --no-warnings=ExperimentalWarning "${CLAUDE_PLUGIN_ROOT}/scripts/session-hook.cjs"

• SubagentStop

  node --no-warnings=ExperimentalWarning "${CLAUDE_PLUGIN_ROOT}/scripts/session-hook.cjs"

• Stop

  node --no-warnings=ExperimentalWarning "${CLAUDE_PLUGIN_ROOT}/scripts/session-hook.cjs"

• PreCompact

  node --no-warnings=ExperimentalWarning "${CLAUDE_PLUGIN_ROOT}/scripts/session-hook.cjs"

• StopFailure

  node --no-warnings=ExperimentalWarning "${CLAUDE_PLUGIN_ROOT}/scripts/session-hook.cjs"

• PostCompact

  node --no-warnings=ExperimentalWarning "${CLAUDE_PLUGIN_ROOT}/scripts/session-hook.cjs"