polytropos

Plan a complex task once on Fable 5 and write the execution kit — PLAN.md, TASKS.md, GUARDRAILS.md, and model-pinned agents — that a cheaper model then executes at near-Fable quality. Use when the user says "plan this with Fable", "architect this", "review this codebase", or escalates a complex task. Not for running a kit; that is /polytropos:execute.

Assess an existing repository for a small, evidence-backed improvement opportunity and produce a scoped findings report and handoff briefs. Use when the user asks where a codebase could be improved, wants an evidence-driven improvement plan, or wants a harness or runtime ecosystem assessed — including multi-harness Polytropos assessments. Not for implementing, activating, enabling providers, or promoting a policy change.

Rank published benchmark entries (Artificial Analysis Intelligence Index), recommend a model+effort per orchestration role given what each harness can actually dispatch, and check that recommendation against this repo's OWN measured routing outcomes. Use when the user asks whether a new/higher model should replace what a role currently runs on, wants a benchmark-informed routing recommendation, or asks "should we upgrade X to Y for this role" — the answer must be checked against the ledger, not just the benchmark.

Reach for this when your context is huge, you're near 700K, cache reads are high, or you're asking should I compact — it explains what this skill can and cannot do, ranks prevent/prune/measure in priority order, and gives the checkpoint move before you compact.

Analyze historical Claude Code usage from local transcripts — spend by model, most expensive sessions, and sessions where a cheaper model would have sufficed. Use when the user asks what they've spent, which model they've been using, or where they could save.

Run one task on the cheapest sufficient model behind a machine-checkable success check, and automatically escalate to a Fable 5.1 subagent (carrying the failure evidence) only if the check fails. Use when the user wants "try it cheap first, fall back to Fable if it doesn't work", an auto-escalating / verify-gated dispatch, or "have Opus attempt it and call Fable if needed".

Decide whether a task is worth running on Fable 5.1 (vs Opus 5.5 / Sonnet 5) and how to run it optimally — effort level and task spec. Use when the user asks "should this run on Fable", "is Fable worth it", or how to get the most out of Fable.

Generate the daily work journal — collect today's AI usage across Claude Code, Copilot CLI, and Codex CLI plus git activity into a digest, then write the narrative, technical, and next-day-plan summaries. Use when the user asks for their work journal, daily summary, "what did I do today", or to plan tomorrow.

Remember durable facts across sessions and recall only the few relevant ones under a strict budget — never bulk-injected. Use when the user says "remember this", asks "what do you remember about…", wants a preference or decision saved, or when a stored fact would materially help the current task.

Benchmark Claude models on a repo's own real work and re-tier them from measured evidence — mined issue-fix pairs or red-validated mutation repairs, sandboxed candidates, four independent oracles. Use when the user asks to benchmark models on their repo, pick a daily driver, or re-tier models for a project. The default is a priced plan that spends nothing; a live run needs a cost ceiling the user confirms in this conversation.

Pick the right Claude model for a task and estimate its cost before running it. Use when the user asks which model to use, whether a task needs Fable/Opus, what a task will cost, or wants a task dispatched to a cheaper model. Args are the task description, optionally prefixed with --api or --sub to force a billing mode.

Install the polytropos statusline (current model, live session cost, context usage, rate-limit burn) into the user's Claude Code settings. Use when the user asks to set up or enable the cost statusline.