Skip to content
Skills Library
AI & agents · Planning

context-optimization

/context-optimization

Use this skill when a task spans 3+ files or crosses repo boundaries and context usage must be managed. Don't use it for single-file edits or read-only queries.

Repository
swarmery ↗
Licence
Apache-2.0
Last repository change
27.10.2026

Assessment and security

100/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content35 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
2
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 27.10.2026, without references to retired models or APIs.

Licence

Apache-2.0, otwarta: można używać i zmieniać, z zachowaniem licencji i informacji o zmianach.

What it does and when to use it

Use this skill when a task spans 3+ files or crosses repo boundaries and context usage must be managed. Don't use it for single-file edits or read-only queries.

Original author description

Use this skill when a task spans 3+ files or crosses repo boundaries and context usage must be managed. Don't use it for single-file edits or read-only queries.

How it works

  1. What it does
  2. When to use it
  3. How to invoke
  4. Worked example
  5. Related
Original SKILL.md headings · 71 lines, 496 words
SKILL.md preview
---
name: context-optimization
version: "1.0.0"
owner: "swarmery-core"
description: "Use this skill when a task spans 3+ files or crosses repo boundaries and context usage must be managed. Don't use it for single-file edits or read-only queries."
disable-model-invocation: true
allowed-tools: Read, Grep, Glob
color: teal
docs:
  status: reviewed
  source_sha: 9d0a19d62c88
  updated: 2026-08-06
---

# Purpose

Plan what to read before reading it. On a task spanning several files or repos,
turn "open everything and hope" into a short ordered context plan: which files,
which line ranges, in which phase, and where to `/clear` between them. It changes
no code — it governs loading so the skills that do the work keep room to think.

# Rules (never violate)

1. Run `codebase-retrieval` before any full-file read; discovery precedes loading.
2. Read with offset/limit — never whole files when one function is the target.
3. Keep files loaded ≤ 3× files actually edited; record the ratio in the plan.
4. Suggest `/clear` at every repo switch and at every phase boundary that frees files.
5. At ≥40% window usage, delegate any isolatable read (one whose product is a
   digest) to a **leaf** subagent at depth-1 — never leaf → leaf.
6. Confidence LOW after discovery means stop and ask, not a plan you half-trust.

# Resources

- Read `resources/planning-procedure.md` before planning: the 7-step procedure,
  the repo/glob map, the step-7 delegation decision table (40% / 60% thresholds),
  the plan template, self-check, escalation triggers, failure modes, and a worked
  two-phase example.

# How to use

## What it does

Produces a context plan: targeted reads in phases, `/clear` decisions with
reasons, and a loaded-versus-edited budget with a confidence rating. It runs
discovery first, so the plan names real files rather than guesses.

## When to use it

- The task touches 3 or more files, in one repo or across several.
- The task crosses a repo boundary — `apps/<mainApp>` plus `<device>`, say.
- Your window is already past 40% and more reads are coming.
- You are about to load a large module tree just to extract a verdict or a list.

Not for: single-file edits (just make them); a read-only question about one
function (`code-search`); or work already running inside a scoped subagent — that
isolation is what this skill would have bought you.

## How to invoke

```
Skill(skill: "core:context-optimization")
Task: fix telemetry latency between the device/edge repo and the main app
Repos: apps/<mainApp>, <device>
```

Invoke it at the start, before any full-file reads. `task_description` is
required; `repos_involved` is inferred from the task and `.claude/project.json`
when omitted.

## Worked example

The invocation above returns a two-phase plan of at most 30 lines. Phase 1 reads
two main-app files by line range — the WebSocket reconnect logic and the SSE
endpoint handler — then marks `/clear before Phase 2: yes`, because the work
switches repos. Phase 2 reads one range in the device sender. Budget: 3 files
loaded, 1 edited, ratio 3:1, confidence HIGH. Had confidence come back LOW, it
would have asked you to narrow the scope instead.

## Related

- `code-search` — when you already know the identifier and want every reference.
- `code-quality` — run it after this skill has narrowed the review set.
- `api-integration` — compose when the task turns on API flows across repos.
Files in this package (2)
  • SKILL.md 3 KB
  • resources/planning-procedure.md 7 KB

Versions

RepositoryScoreLicence
This page · atretyak1985/swarmery100Apache-2.0
dhar174/langgraph_system_generator100MIT
athola/claude-night-market100MIT
AbdullahMalik17/Digital-FTE100MIT
seb1n/awesome-ai-agent-skills100MIT
bcastelino/agent-skills-kit100MIT
viktorbezdek/skillstack100MIT
VoDaiLocz/kilo-kit-mcp90Apache-2.0

Database snapshot: 13.11.2026 · package version 2a37cc9de9b7 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 90token-optimization
    Use the token-optimizer MCP tools to reduce context/token usage when reading, searching, or editing files, or when the context window is filling up. Trigger when reading large files, re-reading files already seen, searching a big/unknown tree, making edits to large files, or…
  • 90advise-with-context
    Answer the user's questions about a context file, plan or task as an advisor, without executing the plan or editing the documents. Use when the user points at a context file or plan and asks questions about it, or asks for an opinion on work another session is driving.
  • 100context-setup
    Install context files from registry. Use when user runs /install-context, says "install context", "setup context", or when context is missing and the user needs to get started.
  • 100agentsop-context-scope-discipline
    Coder-agent working-file budget discipline: keep the editable working set (files you /add into writable context) under ~25k tokens, separate "read" from "edit", delegate breadth to a read-only repo-map, and drop files once edited. Use when an LLM coder-agent edits multiple…
  • 100orchestrate
    Analyzes task dependencies, builds a wave execution plan, and runs specialist agents in parallel and sequential waves to complete a complex multi-agent task. Use when a task spans multiple domains and needs coordinated multi-agent execution.
  • 100context-budget
    Estimate the token cost of reading a file or directory before doing it. Use when asking "how many tokens would this cost", "how big is this file in tokens", "is this file too big to read", before reading large files, or when weighing whether to read something with limited context budget remaining.

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.