CLI proxy that reduces LLM token usage by 60-90%. Declarative YAML filters for Claude Code, Cursor, Copilot, Gemini. rtk alternative in Go.
# Add to your Claude Code skills
git clone https://github.com/edouard-claude/snipYou are an expert at writing declarative YAML filters for snip, a CLI proxy that reduces LLM token consumption by filtering shell output.
filters/*.yaml (embedded in the binary at build time)~/.config/snip/filters/*.yaml (override built-in filters by name)Every filter is a YAML file with this structure:
name: "tool-subcommand" # Required. Unique identifier, used for registry lookup.
version: 1 # Schema version (always 1 for now).
description: "What this filter does" # Human-readable purpose.
match: # Required. When to apply this filter.
command: "tool" # Required. The CLI tool name (e.g., "git", "go", "npm").
subcommand: "sub" # Optional. First non-flag argument (e.g., "test", "log").
exclude_flags: ["-v", "--json"] # Optional. Skip filter if user passes any of these.
require_flags: ["--all"] # Optional. Only apply if user passes ALL of these.
inject: # Optional. Modify command args before execution.
args: ["--json"] # Arguments to append to the command.
defaults: # Flag defaults, only added if flag not already present.
"-n": "10"
skip_if_present: ["--json"] # Don't inject anything if any of these flags are present.
streams: ["stdout", "stderr"] # Optional. Which streams to filter. Default: ["stdout"].
# Use ["stderr"] for tools that output to stderr (e.g., bun test).
# Use ["stdout", "stderr"] to filter both streams merged together.
pipeline: # Required. Ordered list of transformation actions.
- action: "keep_lines"
pattern: "\\S"
- action: "head"
n: 20
on_error: "passthrough" # What to do if the pipeline fails: "passthrough" or "empty".
command is matched exactly against the first token of the shell command.subcommand is matched against the first non-flag argument."-v" matches both and .CLI proxy that filters shell output before it reaches your AI coding assistant's context window. Works with Claude Code, Cursor, Copilot, Gemini CLI, Windsurf, Cline, Codex, Aider, and any tool that runs shell commands.
AI coding agents burn tokens on verbose shell output that adds zero signal. A passing go test produces hundreds of lines the LLM will never use. git log dumps full commit metadata when a one-liner per commit suffices.
snip sits between your AI tool and the shell, filtering output through declarative YAML pipelines -- no compiled filters, no code changes. Write a YAML file, drop it in a folder, done. An extensible rtk alternative built in Go.
snip — Token Savings Report
══════════════════════════════
Commands filtered 128
Tokens saved 2.3M
Avg savings 99.8%
Efficiency Elite
Total time 725.9s
███████████████████░ 100%
14-day trend ▁█▇
Top commands by tokens saved
Command Runs Saved Savings Impact
───────────────────────── ──── ────── ─────── ────────────
go test ./... 8 806.2K 99.8% ████████████
go test ./pkg/... 3 482.9K 99.8% ███████░░░░░
go test ./... -count=1 3 482.0K 99.8% ███████░░░░░
Measured on a real Claude Code session — 128 commands, 2.3M tokens saved.
# Homebrew (macOS/Linux)
brew install edouard-claude/tap/snip
# Or with Go
go install github.com/edouard-claude/snip/cmd/snip@latest
# Then hook into Claude Code
snip init
# That's it. Every shell command Claude runs now goes through snip.
Before — Claude Code sees this (689 tokens):
No comments yet. Be the first to share your thoughts!
-v-verbose"command" or "command:subcommand".args are inserted before any -- separator, otherwise appended.defaults only apply if their flag key is not already present in the user's args.skip_if_present is found, the entire inject block is skipped.| Action | Params | Description |
|--------|--------|-------------|
| keep_lines | pattern (regex) | Keep only lines matching the pattern |
| remove_lines | pattern (regex) | Remove lines matching the pattern |
| head | n (int, default 10), overflow_msg (string, default "+{remaining} more lines") | Keep first N lines |
| tail | n (int, default 10) | Keep last N lines |
| dedup | normalize ([]string of regexes to strip before comparing), top (int, 0=all) | Deduplicate lines, output "text (xN)" for repeats |
| Action | Params | Description |
|--------|--------|-------------|
| truncate_lines | max (int, default 80), ellipsis (string, default "...") | Truncate long lines |
| strip_ansi | (none) | Remove ANSI escape codes |
| compact_path | (none) | Remove directory prefixes from file paths |
| Action | Params | Description |
|--------|--------|-------------|
| regex_extract | pattern (regex with capture groups), format (string using $0, $1, $2...) | Extract data via regex capture groups |
| group_by | pattern (regex with capture group), format (template, default "{{.Key}}: {{.Count}}"), top (int) | Group lines by capture group, count occurrences |
| aggregate | patterns (map of name->regex), format (Go template) | Count matches for named patterns across all input |
| state_machine | states (map of state definitions with keep, until, next) | Stateful line filtering with transitions |
| Action | Params | Description |
|--------|--------|-------------|
| json_extract | fields ([]string), format (template, optional) | Extract fields from JSON input |
| json_schema | max_depth (int, default 3) | Output JSON type schema |
| ndjson_stream | group_by (string field name), format (template with .Key, .Count, .Events) | Process newline-delimited JSON |
| Action | Params | Description |
|--------|--------|-------------|
| format_template | template (Go text/template, required) | Format output using Go template |
format_templateThe template receives:
{{.lines}} - all current lines joined with newlines{{.count}} - number of lines{{.groups}} - map from group_by action (if used earlier in pipeline){{.stats}} - map from aggregate action (if used earlier in pipeline)group_by sets metadata "groups" (map[string]int)aggregate sets metadata "stats" (map[string]int)format_template can access both via {{.groups}} and {{.stats}}keep_lines pattern "\\S" to strip blank lines early.inject to request machine-readable output (e.g., --json, --porcelain) then filter that structured data.exclude_flags to skip filtering when the user explicitly requests a different format.on_error: "passthrough" so raw output is returned if filtering fails.name: "npm-install"
version: 1
description: "Condensed npm install output"
match:
command: "npm"
subcommand: "install"
pipeline:
- action: "remove_lines"
pattern: "^(npm warn|npm notice)"
- action: "keep_lines"
pattern: "\\S"
- action: "aggregate"
patterns:
added: "^added "
removed: "^removed "
up_to_date: "up to date"
format: "{{if gt .up_to_date 0}}up to date{{else}}{{.added}} added, {{.removed}} removed{{end}}"
on_error: "passthrough"
name: "go-test"
version: 1
description: "Condensed go test output with pass/fail summary"
match:
command: "go"
subcommand: "test"
exclude_flags: ["-json", "-v", "-bench", "-run"]
inject:
args: ["-json"]
skip_if_present: ["-json", "-v", "-bench"]
pipeline:
- action: "keep_lines"
pattern: "\\S"
- action: "keep_lines"
pattern: "\"Test\":\""
- action: "keep_lines"
pattern: "\"Action\":\"(pass|fail)\""
- action: "aggregate"
patterns:
passed: '"Action":"pass"'
failed: '"Action":"fail"'
format: "{{if and (eq .passed 0) (eq .failed 0)}}No tests found{{else}}{{.passed}} passed, {{.failed}} failed{{end}}"
on_error: "passthrough"
name: "cargo-test"
version: 1
description: "Condensed cargo test output"
match:
command: "cargo"
subcommand: "test"
pipeline:
- action: "remove_lines"
pattern: "^\\s*(Compiling|Downloading|Downloaded|Updating|Running|Executable)"
- action: "keep_lines"
pattern: "\\S"
- action: "state_machine"
states:
start:
keep: "^(test |running |test result)"
until: "^failures"
next: "failures"
failures:
keep: "."
until: "^$"
next: "done"
- action: "aggregate"
patterns:
pass: "\\.\\.\\. ok$"
fail: "\\.\\.\\. FAILED$"
ignored: "\\.\\.\\. ignored$"
- action: "format_template"
template: "{{.lines}}"
on_error: "passthrough"
inject if so.~/.config/snip/filters/ and running the command through snip.filters/ in the repo and submit a PR.$ go test ./...
ok github.com/edouard-claude/snip/internal/cli 3.728s coverage: 14.4% of statements
ok github.com/edouard-claude/snip/internal/config 2.359s coverage: 65.0% of statements
ok github.com/edouard-claude/snip/internal/display 1.221s coverage: 72.6% of statements
ok github.com/edouard-claude/snip/internal/engine 1.816s coverage: 47.9% of statements
ok github.com/edouard-claude/snip/internal/filter 4.306s coverage: 72.3% of statements
ok github.com/edouard-claude/snip/internal/initcmd 2.981s coverage: 59.1% of statements
ok github.com/edouard-claude/snip/internal/tee 0.614s coverage: 70.6% of statements
ok github.com/edouard-claude/snip/internal/tracking 5.355s coverage: 75.0% of statements
ok github.com/edouard-claude/snip/internal/utils 5.515s coverage: 100.0% of statements
After — snip returns this (16 tokens):
10 passed, 0 failed
That's 97.7% fewer tokens. The LLM gets the same signal — all tests pass — without the noise.
┌─────────────┐ ┌─────────────────┐ ┌──────────────┐ ┌────────────┐
│ Claude Code │────>│ snip intercept │────>│ run command │────>│ filter │
│ runs git │ │ match filter │ │ capture I/O │ │ pipeline │
└─────────────┘ └─────────────────┘ └──────────────┘ └─────┬──────┘
│
┌─────────────────┐ ┌──────────────┐ │
│ Claude Code │<────│ track savings│<──────────┘
│ sees filtered │ │ in SQLite │
└─────────────────┘ └──────────────┘
No filter match? The command passes through unchanged — zero overhead.
| Command | Before | After | Reduction |
|---------|-------:|------:|----------:|
| cargo test | 591 tokens | 5 tokens | 99.2% |
| go test ./... | 689 tokens | 16 tokens | 97.7% |
| git log | 371 tokens | 53 tokens | 85.7% |
| git status | 112 tokens | 16 tokens | 85.7% |
| git diff | 355 tokens | 66 tokens | 81.4% |
Stop wasting tokens on noise. snip gives the LLM the same signal in a fraction of the context window.
brew install edouard-claude/tap/snip
Download the latest binary for your platform from Releases.
# macOS (Apple Silicon)
curl -Lo snip.tar.gz https://github.com/edouard-claude/snip/releases/latest/download/snip_$(curl -s https://api.github.com/repos/edouard-claude/snip/releases/latest | grep tag_name | cut -d'"' -f4 | tr -d v)_darwin_arm64.tar.gz
tar xzf snip.tar.gz && mv snip /usr/local/bin/
go install github.com/edouard-claude/snip/cmd/snip@latest
Or build locally:
git clone https://github.com/edouard-claude/snip.git
cd snip && make install
Requires Go 1.24+ and jq (for the hook script).
snip integrates with every major AI coding assistant. One binary, universal compatibility.
| Tool | Install | Method |
|------|---------|--------|
| Claude Code | snip init | PreToolUse hook |
| Cursor | hooks.json config | beforeShellExecution hook |
| OpenCode | opencode-snip plugin | tool.execute.before hook |
| GitHub Copilot | shell aliases | prefix commands with snip |
| Gemini CLI | shell aliases | prefix commands with snip |
| Codex (OpenAI) | shell aliases | prefix commands with snip |
| Windsurf | shell aliases | prefix commands with snip |
| Cline / Roo Code | shell aliases | prefix commands with snip |
| Aider | shell aliases | prefix commands with snip |
snip init
This installs a PreToolUse hook that transparently rewrites supported commands. Claude Code never sees the substitution -- it receives compressed output as if the original command produced it.
Supported commands: git, go, cargo, npm, npx, yarn, pnpm, docker, kubectl, make, pip, pytest, jest, tsc, eslint, rustc.
snip init --uninstall # remove the hook
Cursor supports hooks since v1.7 via ~/.cursor/hooks.json:
{
"version": 1,
"hooks": {
"beforeShellExecution": [
{ "command": "~/.claude/hooks/snip-rewrite.sh" }
]
}
}
Install the opencode-snip plugin by adding it to your OpenCode config (~/.config/opencode/opencode.json):
{
"$schema": "https://opencode.ai/config.json",
"plugin": ["opencode-snip@latest"]
}
The plugin uses the tool.execute.before hook to automatically prefix all commands with snip. Commands not supported by snip pass through unchanged.
Use shell aliases to route commands through snip:
# Add to ~/.bashrc or ~/.zshrc
alias git="snip git"
alias go="snip go"
alias cargo="snip cargo"
Or instruct the LLM via system prompt to prefix commands with snip.
snip works without any AI tool:
snip git log -10
snip go test ./...
snip gain # token savings report
snip <command> [args] # filter a command
snip gain # full dashboard (summary + sparkline + top commands)
snip gain --daily # daily breakdown
snip gain --weekly # weekly breakdown
snip gain --monthly # monthly breakdown
snip gain --top 10 # top N commands by tokens saved
snip gain --history 20 # last 20 commands
snip gain --json # machine-readable output
snip gain --csv # CSV export
snip -v <command> # verbose mode (show filter details)
snip proxy <command> # force passthrough (no filtering)
snip config # show config
snip init # install Claude Code hook
snip init --uninstall # remove hook
Filters are declarative YAML files. The binary is the engine, filters are data — the two evolve independently.
name: "git-log"
version: 1
description: "Condense git log to hash + message"
match:
command: "git"
subcommand: "log"
exclude_flags: ["--format", "--pretty", "--oneline"]
inject:
args: ["--pretty=format:%h %s (%ar) <%an>", "--no-merges"]
defaults:
"-n": "10"
pipeline:
- action: "keep_lines"
pattern: "\\S"
- action: "truncate_lines"
max: 80
- action: "format_template"
template: "{{.count}} commits:\n{{.lines}}"
on_error: "passthrough"
| Filter | What it does |
|--------|-------------|
| git-status | Categorized status with file counts |
| git-diff | Stat summary, truncated to 30 files |
| git-log | One-line per commit: hash + message + author + date |
| go-test | Pass/fail summary with failure details |
| cargo-test | Pass/fail summary with failure details |
| Action | Description |
|--------|-------------|
| keep_lines | Keep lines matching regex |
| remove_lines | Remove lines matching regex |
| truncate_lines | Truncate lines to max length |
| strip_ansi | Remove ANSI escape codes |
| head / tail | Keep first/last N lines |
| group_by | Group lines by regex capture |
| dedup | Deduplicate with optional normalization |
| json_extract | Extract fields from JSON |
| json_schema | Infer schema from JSON |
| ndjson_stream | Process newline-delimited JSON |
| regex_extract | Extract regex captures |
| state_machine | Multi-state line processing |
| aggregate | Count p