New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Context Crusher

by mohitagw15856 in mohitagw15856/pm-claude-skills

Compress tool outputs, logs, and JSON before they enter the context window — structural compression via a deterministic stdlib script (schema + samples + stats instead of 300 raw rows), no API, no…

MITAuto-check passedAgent Workflows

Install Context Crusher

skills CLI
$ npx skills add mohitagw15856/pm-claude-skills --skill context-crusher -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install mohitagw15856/pm-claude-skills context-crusher --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/mohitagw15856/pm-claude-skills.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/context-crusher .claude/skills/context-crusher && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
context-crusher
GitHub stars
1.4k
Token cost
~1.4k tokens
SKILL.md length
696 words
Files
2 (incl. scripts)
Skills in repo
1,348
Repo updated
First seen
Licence
MIT

At a glance

Compress tool outputs, logs, and JSON before they enter the context window — structural compression via a deterministic stdlib script (schema + samples + stats instead of 300 raw rows), no API, no…

  • Works in 5 steps: Crush between the tool and the context,… → Structural beats semantic for data:… → Errors are sacred: the log crusher's… → …
  • Asked shrink this tool output
  • SKILL.md covers What This Skill Produces, Required Inputs, Programmatic Helper and Framework: The Crush Rules, plus 5 more sections
  • Runs Python scripts from its folder; calls python3

What it does

Context Crusher is an agent skill from mohitagw15856/pm-claude-skills. Compress tool outputs, logs, and JSON before they enter the context window — structural compression via a deterministic stdlib script (schema + samples + stats instead of 300 raw rows), no API, no summarization loss. Use when asked shrink this tool output, my context is full of JSON, compress these logs before analysis, or stop wasting tokens on raw data. Produces the crushed artifact with its token math shown, the crush-or-keep decision rules, and the fetch-the-original escape hatch.

Its SKILL.md is about 1.4k tokens, which your agent loads only when the skill is triggered. The skill folder holds 2 other files, including scripts (for example `scripts/context_crush.py`).

It sits in Agent Workflows, covering Summarization and Context engineering. The repository describes itself as: 1255 professional Agent Skills for Claude, ChatGPT, Gemini, Cursor & Codex — PRDs, postmortems, leases, medical bills, layoffs, go-bags, new countries. Plain markdown, MIT, in… The licence is MIT.

When your agent uses it

  • Asked shrink this tool output
  • My context is full of JSON
  • Compress these logs before analysis
  • Stop wasting tokens on raw data

Example prompts

  • “/context-crusher”

Requirements

  • Python 3

Workflow steps

5 steps, taken from the first numbered list in SKILL.md.

  1. Crush between the tool and the context, not after: the token is spent the moment raw output enters the window — the crush step lives in…
  2. Structural beats semantic for data: summarizing JSON with a model costs tokens, adds latency, and can hallucinate; schema+samples+stats is…
  3. Errors are sacred: the log crusher's contract is that every error/warning line survives regardless of compression — a crush that can lose…
  4. The escape hatch is part of the artifact: every crushed block states where the original lives ("full response in /tmp/response.json…
  5. Know when not to: non-uniform rows where each is signal, data being diffed byte-for-byte, legal/audit content, and anything under ~50…

What it can do on your machine

Read from SKILL.md and the folder at commit 1cbf1f0. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Ships 1 file in scripts/ (Python), which the agent can run.

    Shell commands in SKILL.md call:

    • python3

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    Links to these hosts (documentation or services it may open):

    • github.com

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Context Crusher loads about 1.4k tokens when it runs. Until then it costs about 126 tokens; SKILL.md has 696 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~126
When it runs · the whole SKILL.md, loaded when a task matches
~1.4k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.

SKILL.md

The full file from mohitagw15856/pm-claude-skills at commit 1cbf1f0, republished under its MIT licence (© mohitagw15856). 696 words, ~1,394 tokens.

Download SKILL.mdSave it as .claude/skills/context-crusher/SKILL.md (or your agent's skills folder). This skill also uses 1 other file; get the full folder from GitHub.
name
context-crusher
description
Compress tool outputs, logs, and JSON before they enter the context window — structural compression via a deterministic stdlib script (schema + samples + stats instead of 300 raw rows), no API, no summarization loss. Use when asked shrink this tool output, my context is full of JSON, compress these logs before analysis, or stop wasting tokens on raw data. Produces the crushed artifact with its token math shown, the crush-or-keep decision rules, and the fetch-the-original escape hatch.

Context Crusher Skill

The most expensive tokens in agent work are the ones nobody reads: 300 identical JSON rows when the schema plus three samples would do, a log where one error hides among four hundred heartbeats, a file pasted whole for one relevant section. This skill crushes those structurally — schema + head/tail samples + numeric stats for JSON arrays, dedupe-with-counts plus guaranteed error-line survival for logs, head/tail windowing for text — with a deterministic stdlib script, no model call, no summarization risk. The information that defines meaning survives; the repetition that defines cost doesn't.

What This Skill Produces

  • The crushed artifact — the compressed version, with its token math in the header (~6,000 → ~130 is typical for uniform JSON)
  • The crush decision — what to crush, what to keep raw, and what to link instead of load
  • The escape hatch — every crush names how to fetch the original when a detail turns out to matter
  • The pipeline habit — where in the agent's workflow the crush step belongs (between tool and context, always)

Required Inputs

Ask for these if not provided:

  • The payload — the JSON/log/text (or its path), and roughly how it will be used ("I need the error" vs. "I need every row" are opposite answers)
  • The repetition question — is this data uniform (crushable to schema+stats) or is each row genuinely distinct (crushing loses signal — keep or filter instead)?
  • The journey stage — one-shot analysis (crush hard) vs. data the conversation will keep querying (crush to an index, keep the original fetchable)

Programmatic Helper

bash
python3 scripts/context_crush.py --mode json --file response.json
python3 scripts/context_crush.py --mode log --file build.log --keep 40
cat data.json | python3 scripts/context_crush.py --mode json

Deterministic, stdlib-only, no API. JSON arrays → {count, schema, head samples, tail, numeric min/max/mean} · logs → consecutive-duplicate collapse + first-occurrence dedupe + an always-preserved error/warning section · text → whitespace normalization + head/tail window with an elision marker. Inputs too small to gain are returned unchanged with an honest header.

Framework: The Crush Rules

  1. Crush between the tool and the context, not after: the token is spent the moment raw output enters the window — the crush step lives in the pipeline (tool | crush | context), not in cleanup. Retroactive crushing saves nothing already paid for.
  2. Structural beats semantic for data: summarizing JSON with a model costs tokens, adds latency, and can hallucinate; schema+samples+stats is free, instant, and provably faithful — the numbers are computed, not paraphrased. Save model-summarization for prose, where structure can't do the work.
  3. Errors are sacred: the log crusher's contract is that every error/warning line survives regardless of compression — a crush that can lose the one line that mattered is a corruption, not a compression. Any custom crushing keeps this invariant.
  4. The escape hatch is part of the artifact: every crushed block states where the original lives ("full response in /tmp/response.json — fetch rows by id if needed"), because reversibility is what makes aggressive crushing safe.
  5. Know when not to: non-uniform rows where each is signal, data being diffed byte-for-byte, legal/audit content, and anything under ~50 lines (the crush header costs more than it saves — the script says so itself). Crushing is a default for bulk, not a reflex for everything.
Show full SKILL.md (198 more words)Show less

Output Format

Crushed: [payload] — ~[X] → ~[Y] tokens ([Z]% smaller)

[The crushed artifact, script header included]

Kept raw: [what wasn't crushed and why] · Original: [where it lives, how to fetch] Pipeline note: [where the crush step now sits in this workflow]

Quality Checks

  • The token math appears — before, after, percent
  • JSON crushes carry schema and computed stats, never paraphrased numbers
  • Every error/warning line in a log crush survived
  • The original's location and fetch route are stated
  • Too-small inputs were returned unchanged, honestly

Anti-Patterns

  • Do not summarize data with a model when structure can compress it — paraphrased numbers are hallucination surface
  • Do not crush non-uniform, every-row-is-signal data — filter or keep it
  • Do not drop the escape hatch — irreversible compression turns a saving into a gamble
  • Do not crush after the tokens are spent — the step belongs in the pipeline
  • Do not let the crush eat errors — the invariant outranks the ratio

Based On

The context-compression layer pattern — structural compression of tool outputs before the LLM (as in Headroom) — rebuilt here as a keyless, deterministic, stdlib skill.

Example Trigger Phrases

  • "Shrink this tool output."
  • "My context is full of JSON."
  • "Compress these logs before analysis."
  • "Stop wasting tokens on raw data."

© mohitagw15856, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 1 other file (scripts) in skills/context-crusher of mohitagw15856/pm-claude-skills.

  • SKILL.md
  • scripts/context_crush.py

Open the folder on GitHubat commit 1cbf1f0

Compare with similar skills

Context Crusher next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Context Crusher compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Context Crusher this skillmohitagw15856/pm-claude-skills1.4k—~1.4kAutomated safety check: PassMIT
Self Managed Contextguanyang/open-agent-hub9771 repos~6kAutomated safety check: PassMIT
Context Engineeringmvschwarz/openrig6.6k—~8.2kAutomated safety check: PassApache-2.0
Context Engineeringa5c-ai/babysitter1.8k—~1.8kAutomated safety check: NotesMIT
Context Window Managementaiskillstore/marketplace4333 repos~2.4kAutomated safety check: PassNone
Very Long Text Summarizationcuriositech/some_claude_skills244—~2.2kAutomated safety check: NotesMIT

Similar skills

  • Self Managed Context

    guanyang/open-agent-hub

    This skill should be used when a model gets read-write control over its own live context window instead of a harness-scheduled compaction policy: the context exposed as an editable file the model…

    977 GitHub starsUsed in 1 repo~6k tokens
    Agent WorkflowsAuto-check passed

  • Context Engineering

    mvschwarz/openrig

    A skill your agent uses when designing, reviewing, or debugging how an agent's context window gets filled, pruned, or shared — choosing what loads at boot versus on demand, sizing an install or an…

    6.6k GitHub stars~8.2k tokensUpdated today
    Agent WorkflowsAuto-check passed

  • Context Engineering

    a5c-ai/babysitter

    Context window monitoring and budget management. An agent skill from a5c-ai/babysitter.

    1.8k GitHub stars~1.8k tokensUpdated 23 days ago
    Agent WorkflowsAuto-check: notes

  • Context Window Management

    aiskillstore/marketplace

    Strategies for managing LLM context windows including summarization, trimming, routing, and avoiding context rot

    433 GitHub starsUsed in 3 repos~2.4k tokens
    Agent WorkflowsAuto-check passed

  • Very Long Text Summarization

    curiositech/some_claude_skills

    Summarizes very long texts (books, handbooks, biographies, codebases) using hierarchical multi-pass extraction with cheap model armies.

    244 GitHub stars~2.2k tokensUpdated 1 mo ago
    Writing & ContentAuto-check: notes

  • Teams Channel Post Writer

    daymade/claude-code-skills

    Writes educational Teams channel posts about Claude Code features and best practices, using templates emphasizing concrete examples and underlying principles (e.g.

    1.4k GitHub stars~1.1k tokensUpdated today
    Agent WorkflowsAuto-check passed

More from mohitagw15856/pm-claude-skills

All 1,348 skills in this repo

  • Car Tco

    mohitagw15856/pm-claude-skills

    Compare the total cost of car ownership across buy-new, buy-used, lease, and keep-your-current-car — depreciation, insurance, maintenance ramp, and fuel over a real horizon, not just the monthly…

    1.4k GitHub stars~1.1k tokensUpdated today
    Auto-check passed

  • Cs Health Scorecard

    mohitagw15856/pm-claude-skills

    Build a customer health scorecard for a specific account. An agent skill from mohitagw15856/pm-claude-skills.

    1.4k GitHub stars~2.4k tokensUpdated today
    Auto-check passed

  • Exit Waterfall

    mohitagw15856/pm-claude-skills

    Compute who gets what at each exit price from a cap table — liquidation preferences, conversion points, and where the founders' share collapses.

    1.4k GitHub stars~1.1k tokensUpdated today
    Auto-check passed

  • Feature Prioritisation

    mohitagw15856/pm-claude-skills

    Apply prioritisation frameworks (RICE, MoSCoW, Kano, ICE, Opportunity Scoring) to rank features and backlog items.

    1.4k GitHub stars~2k tokensUpdated today
    Auto-check passed

  • Fire Number

    mohitagw15856/pm-claude-skills

    Compute a financial-independence (FIRE) target and years-to-reach with every assumption labeled as an assumption — plus a sensitivity table instead of a single false-precision answer.

    1.4k GitHub stars~1.1k tokensUpdated today
    Auto-check passed

  • Freelance Rate

    mohitagw15856/pm-claude-skills

    Derive a freelance day/hourly rate backwards from target income, honest billable utilization, overhead, and the self-employment tax premium — the arithmetic that proves a rate is not salary÷2000.

    1.4k GitHub stars~1.2k tokensUpdated today
    Auto-check passed

Categories

Questions about Context Crusher

What does Context Crusher do?

Compress tool outputs, logs, and JSON before they enter the context window — structural compression via a deterministic stdlib script (schema + samples + stats instead of 300 raw rows), no API, no…. Context Crusher is an agent skill from mohitagw15856/pm-claude-skills. Compress tool outputs, logs, and JSON before they enter the context window — structural compression via a deterministic stdlib script (schema + samples + stats instead of 300 raw rows), no API, no summarization loss.

When should I use Context Crusher?

Context Crusher fits situations like: asked shrink this tool output; my context is full of JSON; compress these logs before analysis; stop wasting tokens on raw data.

How do I install Context Crusher in Claude Code?

Run `npx skills add mohitagw15856/pm-claude-skills --skill context-crusher -a claude-code`. Or copy the skill folder (skills/context-crusher in mohitagw15856/pm-claude-skills) into .claude/skills/context-crusher in your project. Claude Code loads it when a task matches its description.

How do I install Context Crusher in Codex?

Run `npx skills add mohitagw15856/pm-claude-skills --skill context-crusher -a codex`. Or copy the skill folder (skills/context-crusher in mohitagw15856/pm-claude-skills) into .agents/skills/context-crusher in your project. Codex loads it when a task matches its description.

Can I use Context Crusher in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add mohitagw15856/pm-claude-skills --skill context-crusher -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/context-crusher, .gemini/skills/context-crusher, .github/skills/context-crusher and .opencode/skills/context-crusher in your project.

What does Context Crusher need to run?

Going by SKILL.md and its folder, Context Crusher needs Python for the scripts in its folder and the command-line tools its instructions call (python3). Our summary lists: Python 3.

Does Context Crusher access the network?

SKILL.md names 1 domain. As links in the text: github.com. This is read from the text; nothing was executed.

Is Context Crusher safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.

What licence does Context Crusher use?

Context Crusher is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Context Crusher use?

About 1.4k tokens (SKILL.md is roughly 5.6k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Context Crusher?

Skills that share tags, products or a category with Context Crusher: Self Managed Context (guanyang/open-agent-hub, 977 stars), Context Engineering (mvschwarz/openrig, 6.6k stars), Context Engineering (a5c-ai/babysitter, 1.8k stars) and Context Window Management (aiskillstore/marketplace, 433 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Context Crusher?

mohitagw15856 (a GitHub user) maintains it in mohitagw15856/pm-claude-skills, which has 1,434 GitHub stars. The repository holds 1,348 skills in this directory. The repository was last updated on October 9, 2026.

Source: mohitagw15856/pm-claude-skills on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.