Estimates token cost of files and directories before reading them into context.
npx skills add https://github.com/vinzenz/workflows --skill read-budget-vinzenz
Please help me install this Agent Skill. Skill: read-budget Source: https://github.com/vinzenz/workflows/tree/main/.agents/skills/read-budget Command: npx skills add https://github.com/vinzenz/workflows --skill read-budget-vinzenz
High-intent search queries and answers about installing and using this skill.
Run read-budget path followed by the file path to get bytes, estimated tokens, a classification, and a read/skim/skip verdict. The estimate uses a calibrated bytes-per-token ratio per file class, not a real tokenizer.
Use read-budget plan with multiple candidate paths. It returns a total cost plus a cheapest-first reading order with a running cumulative_tokens value, so you can stop reading when your budget is exhausted.
Yes, it works outside git repositories. The .gitignore filtering simply does not apply, and the response reports this via the git_repo and gitignore_applied fields.
They are calibrated heuristic estimates, never real tokenizer counts. Ratios were derived from a proxy tokenizer run against real files and biased toward the cheaper end so estimates never dangerously understate cost.
Vendored, generated, or git-ignored directories collapse to a single stat-only byte sum and are never walked file-by-file. Scans are also bounded by a max depth of 20 and 50,000 entries, with any bound hit reported explicitly.
NOMARJScan repositories for malicious patterns and credential exposures using Sigil CLI.
agentsopeEnforce a read-vs-edit split and ~25k token budget for coding agents.
wukangcheng2944Locate relevant RedFi repositories, notes, and relationships across the knowledge vault.
girijashankarjEstimate token usage for files or directories using a word-count heuristic.
