Agent-Skills.md

Agent Skills: Effort Map

Read the current project's current and prior Claude Code or Codex sessions, extract every prompt, score five effort metrics 0-100 with reasons, estimate average tokens and time per session, and render a self-contained light-theme HTML report. Use when the user runs /effort-map or asks how much effort, thought, bug-fixing, UX tuning, architecture care, or copy-paste slop went into a project.

UncategorizedID: diegopacheco/ai-playground/effort-map

Repository

diegopacheco/ai-playground
diegopachecoLicense: Unlicense
218

Install this agent skill to your local

pnpm dlx add-skill https://github.com/diegopacheco/ai-playground/tree/HEAD/pocs/agent-skill-effot-mapper/effort-map

Skill Files

Browse the full folder contents for effort-map.

Download Skill

Loading file tree…

pocs/agent-skill-effot-mapper/effort-map/SKILL.md

Skill Metadata

Name
effort-map
Description
Read the current project's current and prior Claude Code or Codex sessions, extract every prompt, score five effort metrics 0-100 with reasons, estimate average tokens and time per session, and render a self-contained light-theme HTML report. Use when the user runs /effort-map or asks how much effort, thought, bug-fixing, UX tuning, architecture care, or copy-paste slop went into a project.

Effort Map

Score how much real effort went into the current project by reading its agent sessions, then produce effort-DD-MM-YYYY-report.html.

Workflow

  1. Detect the active agent: claude when running in Claude Code, codex when running in Codex.
  2. Collect the sessions for the current project. From this skill directory run: 
    python3 scripts/collect_sessions.py --project "$PWD" \
  --output /tmp/effort-map-sessions.jsonl --stats /tmp/effort-map-stats.json
    Add --agent claude or --agent codex only if automatic detection fails. The command prints token and time stats and writes the normalized transcript.
  3. Treat the current folder as the project boundary. Stop with a clear message when no matching sessions exist.
  4. Read /tmp/effort-map-sessions.jsonl in chunks. Collect every user prompt verbatim (trim noise, drop tool-result echoes and system reminders).
  5. Score each metric from 0 to 100 from the evidence in the transcript, and write one honest sentence of WHY for each:
    • effort — reasoning, planning, and iteration versus one-shotting.
    • bugs — time and turns spent chasing and fixing bugs.
    • experience — tuning UX, polish, and details versus generate-and-done.
    • architecture — care about stack, structure, and design decisions.
    • copy_slop — copying other solutions with no changes or taste (high = more slop).
  6. Merge the stats from /tmp/effort-map-stats.json (avg_tokens, total_tokens, avg_time_minutes, sessions_matched). Do not invent token or time numbers; use the ones the collector computed.
  7. Write /tmp/effort-map-findings.json: 
    {
  "agent": "Claude",
  "project": "project-name",
  "generated": "2026-07-06",
  "sessions_analyzed": 12,
  "prompts": ["first prompt", "second prompt"],
  "stats": { "avg_tokens": 45231, "total_tokens": 542772, "avg_time_minutes": 23.4 },
  "metrics": {
    "effort":       {"score": 72, "why": "..."},
    "bugs":         {"score": 40, "why": "..."},
    "experience":   {"score": 65, "why": "..."},
    "architecture": {"score": 80, "why": "..."},
    "copy_slop":    {"score": 15, "why": "..."}
  }
}
  8. Render the report. From this skill directory run: 
    python3 scripts/render_report.py --input /tmp/effort-map-findings.json --output-dir "$PWD"
  9. Return the absolute report path, the five scores, and the session count.

Scoring Rules

  • Score from evidence only. Base each number on what the transcript shows, not on a guess about the user.
  • Keep every WHY to one specific sentence that names what happened.
  • Remove secrets, credentials, and personal data from prompts before writing them.
  • Never modify the project while scoring. The only file written into the project is the HTML report.