メニュー
Find missed token savings in Claude Code sessions and create filters for unfiltered commands
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Operate Railway infrastructure: sign up for or sign in to a Railway account, create projects, provision services and databases, manage object storage buckets, deploy code, configure environments and variables, manage domains, troubleshoot failures, check status and metrics, set up Railway agent tooling, and query Railway docs. Use this skill whenever the user mentions Railway, signing up, creating an account, registering, logging in, deployments, services, environments, buckets, object storage, build failures, agent setup, MCP, or infrastructure operations, even if they don't say "Railway" explicitly. Also invoke this skill when the user asks to be signed up, registered, or onboarded to Railway: do not refuse — drive them through the unauthed `railway up` flow (deploys + signs up on the fly) or `railway login` (which creates new accounts on the fly).
MUST be used when working with stacked PRs. gh-stack is a CLI tool for managing stacked PR workflows in squash-merge repos (a Graphite alternative). Covers creating stacks, pushing/creating PRs with submit, navigating (up/down/top/bottom), syncing with main, restacking, merging, and agent/CI usage. Triggers on: stacked PR, gh-stack, create stack, submit stack, push stack, restack, sync stack, stack navigation, up down top bottom, PR dependencies, stack visualization, manage PRs, stacked branches.
MUST be used when you need to control smart lights, change light colors, set brightness, turn lights on/off, set light scenes, manage Govee lighting, or run auto-play playlists. Controls Govee smart lights via their cloud API with support for RGB colors, color temperature, brightness, dynamic scenes, DIY scenes, auto-play rotation, and music modes. Triggers on: lights, turn on lights, turn off lights, change light color, set brightness, light scene, govee, smart lights, room lights, mood lighting, dim lights, bright lights, set lights to, light color, warm light, cool light, ambient lighting, curtain lights, TV backlight, auto-play, playlist, DIY scene, rotate scenes.
MUST be used when reviewing GitHub PRs with inline code comments. Uses gh-comment and gh-pr-review CLI extensions for line-specific review comments, batch reviews, thread management, and review submission. Handles the file:line mapping automatically — no manual diff position math needed. Triggers on: review PR, PR review, inline comment, code review, approve PR, request changes, review threads, resolve comment, line comment, submit review, gh comment, gh pr-review.
This skill should be used when the user asks to "create a filter", "write a tokf filter", "add a filter for <tool>", "how do I filter output", or needs guidance on tokf filter step types, templates, pipes, or placement conventions.
MUST be used when creating, updating, searching, assigning, or commenting on Linear issues — anything beyond reading a known ticket ID. `lin` is a human-friendly wrapper over `linctl graphql` that resolves team/project/user names and issue identifiers to IDs internally, so you stop fumbling syntax. Covers child issues with parent links, assigning to other users, fuzzy user/team/project lookup, filtered search, the learned disambiguation cache, and the raw GraphQL escape hatch. Triggers on: linear ticket, create issue, child ticket, sub-issue, assign ticket, assign to, link parent, search linear, find tickets, linear comment, update issue, lin cli, linctl, who is on linear, ambiguous name linear.
| name | tokf-discover |
| description | Find missed token savings in Claude Code sessions and create filters for unfiltered commands |
| user_invocable | true |
Use this skill to analyze Claude Code sessions and find commands that are running without tokf filtering, wasting tokens on verbose output.
Run tokf discover in the project directory to scan recent sessions:
tokf discover
--all — scan all projects, not just the current one--since 7d — only scan sessions from the last 7 days (also 24h, 30m)--limit 0 — show all results (default: top 20)--json — output as JSON for programmatic use--session <path> — scan a specific session file--project <path> — scan sessions for a specific project pathThe output shows:
tokf discover to identify top savings opportunitiestokf hook install to set up automatic filtering/tokf-filter skill to create a custom filtertokf discover after changes to verify improvementIf tokf discover shows commands with no matching filter, create one:
# See what a filter would look like
tokf which "the-command --args"
# Use the tokf-filter skill to create a proper filter
# /tokf-filter
Use --json for integration with other tools:
tokf discover --json | jq '.results[] | select(.estimated_savings > 1000)'
The JSON schema includes:
sessions_scanned — number of JSONL files processedtotal_commands — all Bash commands foundalready_filtered — commands already using tokffilterable_commands — commands with available filtersno_filter_commands — commands with no matching filterestimated_total_savings — total estimated token savingsresults[] — per-command breakdown sorted by savings