Skip to content

show-command.md

Latest commit

 

History

History
97 lines (68 loc) · 3.52 KB

show-command.md

File metadata and controls

97 lines (68 loc) · 3.52 KB
Raw

Show Command

Print collected analytics for one spec — or for the whole project. Pure local rendering.

Usage

Inside Claude Code

/speckit.analytics.show <spec-name>
/speckit.analytics.show --all
/speckit.analytics.show

The slash command resolves to a single Bash invocation of the underlying script, so the LLM only spends a handful of tokens on dispatch.

Tip — bypass the LLM entirely. Prefix the script with ! in Claude Code; the harness runs the command directly in your shell without consulting the model:

! ./.specify/extensions/analytics/scripts/bash/show.sh <spec-name>

GitHub Copilot, Cursor Chat, Continue and similar assistants do not have an equivalent of ! — every message there routes through the LLM. For a zero-token run with those tools, just open your IDE's integrated terminal (without the chat panel) and run the script directly.

Outside Claude Code (terminal, CI, scripts)

Run the bash script directly from the project root (the directory containing .specify/):

./.specify/extensions/analytics/scripts/bash/show.sh <spec-name>
./.specify/extensions/analytics/scripts/bash/show.sh --all
./.specify/extensions/analytics/scripts/bash/show.sh

If you call it from a subdirectory, point it at the project root explicitly:

SPECKIT_PROJECT_ROOT="/abs/path/to/project" \
  /abs/path/to/project/.specify/extensions/analytics/scripts/bash/show.sh <spec-name>

Zero setup, zero token cost — useful in CI, cron jobs, or any non-interactive context.

What gets printed

For each spec, up to three sections are rendered (a section is omitted if it has no data):

  1. ⏱ Execution Timer — one line per recorded command, with the cumulated real time.
  2. 👤 Human Time Estimates (per command) — junior and senior estimates per command, with totals.
  3. 📋 Task Projections — junior and senior estimates per task, with totals.

When more than one spec is rendered, a Σ Project totals footer is appended with:

  • Real execution time across the project.
  • Total estimated human time (junior / senior).
  • Time saved against both junior and senior baselines.

Sample output

📊 spec-kit-analytics

📁 Spec: demo-feat
────────────────────────────────────────────────────────────
⏱  Execution Timer
  • plan "demo"  →  1m 13s  [Claude Opus 4.7]
  • tasks "demo"  →  12s    [Claude Opus 4.7]
  Σ total: 1m 25s

👤 Human Time Estimates (per command)
  • plan   →  junior 4h, senior 1h 30m
  • tasks  →  junior 4h, senior 1h 30m
  Σ junior: 8h   senior: 3h

📋 Task Projections
  • T001  Build auth   →  junior 8h, senior 3h
  • T002  Write tests  →  junior 4h, senior 1h 30m
  Σ junior: 12h   senior: 4h 30m

Data sources

Both Markdown and JSON storage formats are supported transparently. The renderer reads whatever exists under:

<spec-folder>/analytics/time/commands.{md,json}
<spec-folder>/analytics/time/commands/<name>.{md,json}        (splitByCommands)
<spec-folder>/analytics/time/humanEstimates/commands.{md,json}
<spec-folder>/analytics/time/humanEstimates/<name>.{md,json}  (splitByCommands)
<spec-folder>/analytics/time/humanEstimates/tasks.{md,json}

Notes

  • Read-only: the show command never writes to disk and never modifies analytics files.
  • If no analytics exist yet, a friendly hint is printed.
  • If a spec name is passed but does not match any folder, an empty result is rendered (no error).