NodeJSmith/cm-get-token-insights
skills-memory/cm-get-token-insights/SKILL.md
Use when the user asks about Claude token usage, wants to see how much they are spending on Claude, understand cache hit rates, review Claude Code workflow patterns, or get cost optimization recommendations.
$ install --global
skillsauthnpx skillsauth add NodeJSmith/Claudefiles cm-get-token-insightsInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: Apr 27, 2026, 2:29 AM109.1s1 file scanned
SKILL.md
- name:
- cm-get-token-insights
- user-invocable:
- true
- description:
- >
Get Token Insights
Parse JSONL conversation files from ~/.claude/projects/*/ into per-turn analytics tables, then analyze both cost-optimization opportunities and Claude Code workflow patterns (skills, agents, hooks).
Step 1: Ingest
cm-ingest-token-data
First run processes all files (~100s for ~2500 files) — warn the user about the wait before running. Incremental runs complete in under 5s. The script populates analytics tables, deploys an interactive dashboard to ~/.claude-memory/dashboard.html (built from templates/dashboard.html), and prints a slim JSON blob to stdout (full data goes to dashboard only).
If the script exits non-zero, report the error and stop.
Step 1.5: Claude Code Feature Enrichment
After parsing the JSON stdout from Step 1, construct a personalized prompt for a claude-code-guide agent using the actual data — not generic descriptions. For each of the top 3 insights (by waste_usd), include verbatim: the finding text, root_cause text, waste_usd value, solution.action, and solution.detail. Also include the specific project names, counts, and numbers mentioned in the insight (e.g. "meta-ads-cli: 75 cliffs across 53 sessions") so the agent's response is grounded in the user's real usage patterns.
Spawn the agent with subagent_type: "claude-code-guide" in foreground (do not use run_in_background). Wait for the agent to return before proceeding to Step 2. Weave its suggestions into the analysis in Step 2.
Step 2: Analyze
Capture the JSON stdout from Step 1 as the analysis input. Analyze across four areas:
- Cost optimization — top insights by dollar waste: finding, root cause, concrete solution (show exact CLAUDE.md rule text if suggested), estimated savings. Include relevant Claude Code feature suggestions from Step 1.5. Add a top-line summary (total spend, session count, date range, avg cost/session) and model economics (cost by model, savings from switching routine tasks to cheaper models).
- Workflow patterns — skill usage and error rates, agent delegation patterns (subagent types, model overrides, unnecessary general-purpose usage), hook performance (slowest by total runtime, high error rates).
- Week-on-week trends — if the
trendsobject is non-empty, compare current vs prior window: improved/regressed metrics with likely causes, new/retired skills and hooks, hook latency deltas. Skip iftrendsis empty. - Project cost ranking — top 3 projects by spend; for the most expensive, identify what drives the cost.
Structure the analysis naturally based on what the data shows — don't force empty sections. Ask the user if they want to dive deeper into any specific project, skill, or insight.
Step 3: Open Dashboard
python3 -c "import webbrowser, pathlib; webbrowser.open((pathlib.Path.home() / '.claude-memory' / 'dashboard.html').as_uri())"
Note the dashboard is available for deeper exploration — Section 6 (Claude Code Ecosystem) has the new skill, agent, and hook charts.
Related Skills
NodeJSmith/mine-document
development
VerifiedTrustedCommunity
Use when the user says: "document how X works", "write up how this works", "durable explanation", "explain this for the docs", "document this subsystem". Writes a durable, architectural-altitude explanation that survives code churn.
1SKILL.mdUpdated Jul 6, 2026
NodeJSmith/cass-resume
development
VerifiedTrustedCommunity
Use when picking up a fresh session after /clear, a stop, or an unanswered AskUserQuestion. Reconstructs the prior session's intent from its transcript tail and surfaces any unresolved decision; user-invoked only — for a hand-written end-of-day handoff use /mine-good-morning instead.
1SKILL.mdUpdated Jul 4, 2026
NodeJSmith/cass-recall
development
VerifiedTrustedCommunity
Use when the user says: "what did we discuss", "continue where we left off", "remember when", "as I mentioned", "you suggested", "we decided", "search my conversations", "find the conversation where", "what did we work on", or uses implicit signals like past-tense references, possessives without context, or assumptive questions. Direct search over past Claude Code sessions via cass.
1SKILL.mdUpdated Jul 4, 2026
NodeJSmith/cass-context
tools
VerifiedTrustedCommunity
Use when the user says: "what context do I have", "relevant history for this task", "what have we done related to this". Also usable proactively when starting work that likely has prior history. Assembles a structured context brief from past session history via cass, scoped to the current task and workspace.
1SKILL.mdUpdated Jul 4, 2026