agentleFS
Sign inStart free
Golden-PromiseGolden-Promise/codex-skill-library/skills/skill-context-keeper/SKILL.md

skill-context-keeper

Skill1 starsChanged 8 months agoMITView on GitHub

Use when the user needs to refresh or compress trusted root-task state for an ongoing coding task without creating subtask-local state, workflow gates, or final handoffs.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 4 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use this skill when the thread needs root-state refresh or compression for an artifact such as .agent-state/root/TASK_STATE.md, not …

    line 11
  2. Distinctive:

    Keep the output focused on root-state refresh and compression, not future workflow control.

    line 31
  3. Distinctive:

    Do not create subtask-local state.

    line 35
  4. Distinctive:

    Do not run phase gates.

    line 36
  5. Distinctive:

    Do not generate final handoffs.

    line 37

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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

Skill Context Keeper

Overview

Maintain trusted root-task state for long-running coding tasks. Use this skill when the thread needs root-state refresh or compression for an artifact such as .agent-state/root/TASK_STATE.md, not subtask-local state, packet compression, phase gates, or final transfer notes.

Use This Skill When

  • resuming a task after an interruption, stale summary, or context loss at the root-task level
  • rebuilding the last known root task state before new work continues
  • refreshing verified facts, decisions, assumptions, or next actions in .agent-state/root/TASK_STATE.md
  • compressing stale root-task detail into archive-minded notes instead of letting the active summary grow indefinitely
  • reconciling root-state changes since the last stable checkpoint

Do Not Use This Skill When

  • the user needs a bounded child task with its own local state
  • the next step only needs a packet-sized context object
  • the user needs workflow phases, checkpoints, or exit criteria
  • the thread needs a final pause summary or transfer handoff
  • the request is asking one package to coordinate the full continuity suite

Core Rules

  1. Keep the output focused on root-state refresh and compression, not future workflow control.
  2. Distinguish verified facts from assumptions and from decisions already made.
  3. Assume downstream path examples such as .agent-state/root/TASK_STATE.md unless the user specifies another target.
  4. Move stale detail into compression or archive notes instead of inflating the active summary.
  5. Do not create subtask-local state.
  6. Do not run phase gates.
  7. Do not generate final handoffs.

References

  • README.md and README.zh-CN.md: package overview, installation, and boundary guidance
  • references/use-cases.md and references/use-cases.zh-CN.md: reader-facing trigger examples
  • references/prompt-templates.en.md and references/prompt-templates.zh-CN.md: reusable root-state refresh prompts
  • assets/TASK_STATE.template.md: compact root task-state artifact template

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