agentleFS
Sign inStart free
Wsh7AshWsh7Ash/skillmcp/skills/context-window-management/SKILL.md

context-window-management

Skill0 starsChanged 14 days agoMITView on GitHub

Spend context on the procedure you will follow, not on everything the server could return. Use when a session accumulates tool output and the budget for reasoning is shrinking

8 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 6 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Prefer summaries and counts over full documents when deciding

    line 21
  2. Distinctive:

    no document is fetched twice

    line 32
  3. Distinctive:

    only the acted-on documents are kept in full

    line 62

“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Context Window Management

Spend context on the procedure you will follow, not on everything the server could return

Use when: a session accumulates tool output and the budget for reasoning is shrinking

Avoid when: the session is short and every returned document is used immediately

Procedure

  1. Track roughly how much text each tool call returned
  2. Prefer summaries and counts over full documents when deciding
  3. Fetch full text only for the item you will act on
  4. Drop or summarise earlier output that no longer affects the task
  5. Re-fetch a needed detail by id rather than keeping it in context
  6. Record the identifiers you will need later, not the bodies

Signals

  • the working notes contain identifiers rather than full documents
  • each fetch is followed by an action
  • summaries replace full listings once a decision is made
  • no document is fetched twice

Patterns

1. Identifiers not bodies

keep skillmcp slug list, not 50 SKILL.md files

2. Summarise

10 matches -> 2 candidates and why

3. Re-fetch by id

get_skill(slug) when the detail is needed again

Pitfalls

  • Fetching a whole pack to answer one question
  • Keeping every response verbatim and running out of room for reasoning
  • Re-fetching the same document because the identifier was not recorded

Done when

  • only the acted-on documents are kept in full
  • identifiers are recorded so any document can be re-fetched

References

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free