Aee Report Savings Skill

AEE Report Savings

Report token savings measured by the optional token-economy tools after a large Spec Kit workflow, after implementation, or when asked how many tokens were saved.

User input

$ARGUMENTS

Accept an optional reporting window in hours (default 168).

Measurement rules

• Count only RTK and Headroom as token-saving tools.
• RTK saves shell-output tokens. Measure with rtk gain.
• Headroom saves context and tool-output tokens. Measure with headroom perf --hours <N>.
• Do not count token-router, Ollama, subagents, or state externalization as token savings. They are workflow tools, not savings channels.
• Never estimate. If a measurement source is missing or unparseable, report n/a with the reason.
• Keep RTK and Headroom separate. Add a combined total only when both values are measured in the same report. Never double-count.

Execution

Run the policy helper:

python .specify/extensions/aee/scripts/python/aee_token_economy.py report --hours "<N>"

The helper degrades gracefully: a missing tool yields n/a with a reason, never a zero and never an estimate.

Report format

The helper prints JSON; report its values verbatim, per channel:

{
  "rtk": {"used": true, "saved_tokens": 1234, "source": "rtk gain --format json"},
  "headroom": {"used": false, "saved_tokens": "n/a", "reason": "headroom not installed"},
  "combined_saved_tokens": "n/a",
  "policy": "…"
}

used is a JSON boolean (true/false, never yes|no|n/a); saved_tokens is a measured number or the string "n/a" with a reason. combined_saved_tokens is the sum only when both channels measured, otherwise "n/a".

Save the report under .specify/extensions/aee/reports/ when it belongs to a feature workflow. Include the commands used as evidence. Do not print raw tool logs that contain sensitive file paths or secrets.

Guardrails

• A savings figure is a measurement claim: ground it in the tool output, or mark it n/a.
• Do not quote vendor marketing figures for any tool. Only measured values from this environment count.