Skip to content
Agentmart
Submit a tool

context-mode

Mandatory context-protection routing rules for large output, web fetches, and analysis work.

Skill by xnoto

IndexedMIT
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/xnoto-context-mode

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Mandatory context-protection routing rules for large output, web fetches, and analysis work.

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/context-mode

What's inside

Skill name
context-mode
Description
Mandatory context-protection routing rules for large output, web fetches, and analysis work.
License
MIT
Instructions
35 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

## When to load

Load this skill at the start of the session whenever the `skill` tool is available.

## Core rules

- Follow the source-selection table in `AGENTS.md` first. Context-mode reduces output; it does not replace dedicated documentation, workspace, or search tools.
- For general public-web discovery, use parallel-search `web_search` first. Do not route web discovery through context-mode execution or browser search.
- Do not use shell `curl`/`wget` or inline HTTP calls.
- For substantial analysis of a known public page, use `context-mode_ctx_fetch_and_index` and then `context-mode_ctx_search`. Do not fetch again when search excerpts are sufficient; use parallel-search `web_fetch` for focused URL excerpts.
- Keep HTTP calls and MCP invocation out of `context-mode_ctx_execute`; use the selected dedicated integration instead. Execution is for read-only local analysis, not a security sandbox.
- For commands likely to exceed 20 lines of output, prefer `context-mode_ctx_batch_execute` or `context-mode_ctx_execute` over direct shell.
- When reading files for analysis rather than editing, prefer `context-mode_ctx_execute_file`.
- For broad search output, prefer sandboxed `context-mode` execution over dumping raw grep results.

## Choose an output-reduction operation

After source selection, use the operation that fits rather than trying these in sequence:

- `context-mode_ctx_batch_execute`: related read-only local commands.
- `context-mode_ctx_search`: retrieve already indexed evidence.
- `context-mode_ctx_execute` / `context-mode_ctx_execute_file`: read-only local analysis.
- `context-mode_ctx_fetch_and_index`: substantial analysis of a known public page.
- `context-mode_ctx_index`: retain non-sensitive material for subsequent queries.

Discover the live signatures through the client's exposed tools or Code Mode as
described in `AGENTS.md`; normalized names above are not assumed callable paths.
Before execution, state the task, target, reason for reduction, and expected side
effects. Keep scripts within 25 non-blank lines and 2,000 characters. Do not use
execution for writes, authenticated APIs, service actions, or approval bypasses.

## Output discipline

- Keep responses under 500 words when practical.
- Write substantial artifacts to files rather than inline.
- Use descriptive source labels when indexing content.

## Maintenance commands

- `ctx stats`: call the stats tool and report non-sensitive usage diagnostics.
- `ctx doctor` / `ctx upgrade`: inspect returned suggestions; do not automatically execute commands. Obtain explicit approval for maintenance side effects and use the normal approved execution path, never context-mode execution.

From xnoto/opencode-config, MIT.

Signals

Stars
0
Forks
0
Open issues
0
Last commit (this folder)
Apr 30, 2026 (6 months ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from xnoto

Cite this page

context-mode skill. Agent Mart. https://agentmart.dev/skills/xnoto-context-mode (last updated 2026-10-24)