Token DashboardSAFE

Find the ghost tokens. Fix them. Survive compaction. Avoid context quality decay.

The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.

---
name: token-dashboard
description: Answers what a past session cost ("the Garden session yesterday", a session id) with a priced number, and opens the Token Optimizer dashboard when asked to see it.
---

# Token Optimizer Dashboard

Answers what a session cost from the user's own words, and opens an up-to-date dashboard of context usage, quality, session history and skills. Open the dashboard only when the user asks to see it; a cost question never opens a browser.

## Instructions

1. **Resolve runtime and measure.py path**:
```bash
RUNTIME="${TOKEN_OPTIMIZER_RUNTIME:-}"
if [ -z "$RUNTIME" ]; then
  if [ -n "$CLAUDE_PLUGIN_ROOT" ] || [ -n "$CLAUDE_PLUGIN_DATA" ]; then
    RUNTIME="claude"
  elif [ -n "$OPENCODE" ] || [ -n "$OPENCODE_BIN" ] || [ -n "$OPENCODE_CONFIG_DIR" ] || [ -n "$OPENCODE_CONFIG" ]; then
    RUNTIME="opencode"
  elif [ -n "$CODEX_HOME" ]; then
    RUNTIME="codex"
  # Explicit config-dir env of another harness: same tier as Python's
  # detect_runtime() (above CLAUDECODE, which every Claude Code child inherits).
  elif [ -n "$HERMES_HOME" ]; then
    RUNTIME="hermes"
  elif [ -n "$COPILOT_HOME" ] || [ -n "$TOKEN_OPTIMIZER_COPILOT_HOME" ]; then
    RUNTIME="copilot"
  elif [ -n "$TOKEN_OPTIMIZER_CURSOR_HOME" ]; then
    RUNTIME="cursor"
  elif [ -n "$TOKEN_OPTIMIZER_ANTIGRAVITY_HOME" ]; then
    RUNTIME="antigravity"
  elif [ -n "$GROK_HOME" ] || [ -n "$TOKEN_OPTIMIZER_GROK_HOME" ]; then
    RUNTIME="grok"
  elif [ -n "$CLAUDECODE" ] || [ -n "$CLAUDE_CODE_ENTRYPOINT" ] || [ -n "$CLAUDE_CODE_SESSION_ID" ]; then
    RUNTIME="claude"
  elif [ -d "$HOME/.config/opencode" ] && [ ! -d "$HOME/.codex" ]; then
    RUNTIME="opencode"
  elif [ -d "$HOME/.codex" ]; then
    RUNTIME="codex"
  else
    RUNTIME="claude"
  fi
fi

# Resolve measure.py to the NEWEST installed copy across channels so a stale
# plugin-cache copy never shadows a fresh install. find -L follows the
# install.sh symlink under ~/.claude/skills; cd -P resolves it before reading each
# copy's plugin.json for its version. find (not bare globs) never errors under zsh.
MEASURE_PY=""; _best_ver=""
while IFS= read -r _cand; do
  [ -f "$_cand" ] || continue
  _root="$(cd -P -- "$(dirname -- "$_cand")/../../.." 2>/dev/null && pwd)"
  _ver="$(sed -n 's/.*"version"[[:space:]]*:[[:space:]]*"\([^"]*\)".*/\1/p' "$_root/.claude-plugin/plugin.json" 2>/dev/null | head -1)"
  [ -n "$_ver" ] || _ver="0.0.0"
  if [ -z "$_best_ver" ] || [ "$(printf '%s\n%s\n' "$_ver" "$_best_ver" | sort -t. -k1,1n -k2,2n -k3,3n -k4,4n | tail -n1)" = "$_ver" ]; then
    _best_ver="$_ver"; MEASURE_PY="$_cand"
  fi
done <<EOF
$(find -L "$HOME/.claude/skills" "$HOME/.claude/plugins/cache" "$HOME/.claude/token-optimizer" "$HOME/.codex/skills" "$HOME/.codex/plugins/cache" "$HOME/.config/opencode/plugins/cache" "$HOME/.config/opencode/plugins" -type f -name measure.py -path '*token-optimizer*/scripts/measure.py' 2>/dev/null)
EOF
if [ -z "$MEASURE_PY" ]; then echo "[Error] measure.py not found. Is Token Optimizer installed?"; exit 1; fi
export TOKEN_OPTIMIZER_RUNTIME="$RUNTIME"
```

2. **A cost question** ("what did the Garden session yesterday cost?", "how much were my TR sync sessions?", "cost of session <id>"): run this instead of steps 3-4.
```bash
python3 "$MEASURE_PY" session-cost <the user's words> --json
```
   - Pass the user's own words (`garden yesterday`, `tr sync`, an id or its first 8 characters). Names match topic, slug and project, case-insensitively, and all of them must match; `today`, `yesterday` and `YYYY-MM-DD` set the day. Add `--days N` for "this week" (default 30), `--limit N` to list more.
   - Answer from `sessions[]`: date, topic, `cost_usd.total`, the main-thread and subagent split when subagents ran, `cost_usd.on_date` when a day was asked and the session ran over several days, and `partial_reason` when set. Several matches: give `total_cost_usd` and list them, asking which one only when it changes the answer.
   - State `basis` in one line: API list prices, so on a subscription it is what the work would have cost, not a bill.
   - No match: say so, show `closest[]`, and pass on `note` (a session is collected when it ends).
   - Never query trends.db or price tokens yourself: this is the dashboard's own number, read without writing.

3. **Collect and open**:
```bash
# TOKEN_OPTIMIZER_INTERACTIVE=1 marks this as a user-initiated open, so the 20s
# hook budget does not cut off a heavy rebuild on a large history. The user is
# waiting for this open, so it runs unbounded.
TOKEN_OPTIMIZER_INTERACTIVE=1 python3 "$MEASURE_PY" collect --quiet && \
  TOKEN_OPTIMIZER_INTERACTIVE=1 python3 "$MEASURE_PY" dashboard
```

This collects the latest session data into the trends database, regenerates the dashboard HTML, and opens it in your default browser.

4. **Tell the user** the dashboard is open. Probe the daemon before mentioning any URL:
   - Probe daemon: `python3 "$MEASURE_PY" daemon-status 2>/dev/null`
   - If DAEMON_RUNNING: lead with `URL: http://localhost:24842/token-optimizer` (bookmarkable, auto-updates), then mention the file fallback.
   - If DAEMON_NOT_RUNNING: do not print the `localhost:24842` URL. Tell the user the dashboard opened as a file, and suggest `python3 $MEASURE_PY setup-daemon` (macOS and Windows) if they want a bookmarkable URL.
   - File path: never hardcode it. It is install-dependent. Cite the `  Dashboard: ` line from the step-3 output.
   - For Codex when DAEMON_NOT_RUNNING: do not imply the Claude daemon is required; the generated file works, and Stop hooks refresh it when balanced hooks are installed.

Written for and checked on Claude Opus 5.5 and Sonnet 5.5, and GPT-5.6 on Codex.

SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

What the source does

Filesystem

none-observed

Network

none-observed

Shell

none-observed

Dependencies

pinned

Secrets in source

none-found

Findings (0)

No findings outside the package's declared scope.

Gates applied: no_behavioural_pass.

[![ModelsAtlas audit: Token Dashboard](https://modelsatlas.com/skills/alexgreensh/token-dashboard/badge.svg)](https://modelsatlas.com/skills/alexgreensh/token-dashboard/)

Not your listing, or something out of date? Claim or update your listing.