session-report

name	session-report
description	Generate an explorable HTML report of Claude Code session usage (tokens, cache, subagents, skills, expensive prompts) from ~/.claude/projects transcripts. Use at end-of-session, end-of-week, or when investigating token spend / cache-hit anomalies / which subagent or skill is eating budget.

Session Report — Claude Code Usage Retrospective

Produce a self-contained HTML report of Claude Code usage and save it to the current working directory.

Source bundle

The bundled analyzer + template live under the marketplace plugin at: C:\Users\kalis\.claude\plugins\marketplaces\claude-plugins-official\plugins\session-report\skills\session-report\

• analyze-sessions.mjs — extracts data from ~/.claude/projects/* transcripts
• template.html — interactive report shell

These are NOT copied into FLOSS/skill-corpus to avoid drift from upstream. This skill points at them directly.

Steps

1. Get data. Run the bundled analyzer (default window: last 7 days; honor a different range if the user passed one — 24h, 30d, or all):

   node "C:\Users\kalis\.claude\plugins\marketplaces\claude-plugins-official\plugins\session-report\skills\session-report\analyze-sessions.mjs" --json --since 7d > /tmp/session-report.json
   

   For all-time, omit --since. On Windows substitute a writable temp dir (e.g. $env:TEMP\session-report.json).

2. Read /tmp/session-report.json. Skim overall, by_project, by_subagent_type, by_skill, cache_breaks, top_prompts.

3. Copy the template to a timestamped HTML in the current working directory:

   cp "C:\Users\kalis\.claude\plugins\marketplaces\claude-plugins-official\plugins\session-report\skills\session-report\template.html" "./session-report-$(date +%Y%m%d-%H%M).html"
   

4. Edit the output file (use Edit, not Write — preserve the template's JS/CSS):

   • Replace contents of <script id="report-data" type="application/json"> with the full JSON from step 1.
   • Fill <!-- AGENT: anomalies --> with 3–5 one-line findings. Express figures as % of total tokens wherever possible (total = overall.input_tokens.total + overall.output_tokens). One line per finding, exact markup:

     <div class="take bad"><div class="fig">41.2%</div><div class="txt"><b>cc-monitor</b> consumed 41% of the week across just 3 sessions</div></div>
     

     Classes: .take bad red, .take good green, .take info blue. The .fig is one short number (%, count, or multiplier like 12×). The .txt is one plain-English sentence naming the project/skill/prompt; wrap the subject in <b>. Look for: a project or skill eating a disproportionate share, cache-hit <85%, a single prompt >2% of total, subagent types averaging >1M tokens/call, cache breaks clustering.
   • Fill the <!-- AGENT: optimizations --> block (bottom of page) with 1–4 <div class="callout"> suggestions tied to specific rows.
   • Do not restructure existing sections.

5. Report the saved file path. Do not open it.

FLOSSI0ULLK-specific anomaly checks

When generating the narrative, also flag:

• Heartbeat-loop spend > 30% of weekly total → potential STOP-gate drift; see FLOSS/docs/specs/heartbeat-runtime-budget.spec.md.
• A single voter/model > 50% of consensus-gateway spend → diversity policy violation; check packages/metacoordinator_mcp/voter_registry.json.
• Synthesis drafts staged > generated → backpressure should kick in; check FLOSS_SYNTHESIS_STAGING_CAP.
• A single skill from skill-corpus/ >15% of total → consider whether the skill is too heavy-context or duplicated across harnesses.

Notes

• The template is the source of interactivity. Your job is data + narrative, not markup.
• Keep commentary terse and specific — reference actual project names, numbers, timestamps from the JSON.
• top_prompts already includes subagent tokens and rolls task-notification continuations into the originating prompt.
• If JSON > 2MB, trim top_prompts and cache_breaks to 100 entries before embedding.
• Cross-harness note: this skill targets the Claude Code transcript layout under ~/.claude/projects/. Codex/Gemini/OpenCode transcripts live elsewhere and need their own session-report variants — track as a future skill if the user runs into the same anomaly-finding need on those harnesses.

Provenance

Original skill: claude-plugins-official/session-report. Adapted for FLOSSI0ULLK: added the heartbeat/voter/synthesis/skill anomaly checks tied to the FLOSSI0ULLK runtime surfaces, noted cross-harness boundary.