Alphakdeps.io is experimental and evolving — the registry and packages may change or reset.
Harnesskdeps asset

kdeps-harness-refine-system

v1.0.0

kdeps harness: refine-system

set harness / shelf context

PurposeThe text kdeps sends the model in agent mode: the sections of the system prompt, and the prompts behind compaction, goals, judges, prompt refinement and nudges.

LLM context: ~216 tokens when loaded (prompt or tool definition)

Where it runs

Part of the harness set, used in workflow and agent mode. The text kdeps sends the model in agent mode: the sections of the system prompt, and the prompts behind compaction, goals, judges, prompt refinement and nudges.

  1. you type a prompt
  2. the agent loop builds the request
  3. the harness sections become the system promptharness/refine-system
  4. the model answers
  5. the loop runs tools or replies

Loaded by: kdeps (agent loop), /settings, kdeps run, kdeps konfig export, kdeps knowledge build, kdeps konfig import

In the code: kdeps (agent loop)agent.Newagent.harnessText

Install

A kdeps asset: installed with the first-run profile that includes its set, or the first time a command needs it, and updated without a new release.

kdeps install harness/refine-system

Pin a version with kdeps install harness/refine-system@<version>.

Contents

name
refine-system
kind
standalone
You rewrite a user's request so it is clearer for an AI agent to act on.

The conversation history above (if any) is context ONLY, for resolving what the
request refers to. Do not answer it, continue it, or rewrite any earlier turn.

Reply with ONLY the rewritten request as plain text. No preamble, no quotes, no explanation.

Rules:
- Preserve the user's intent and every concrete detail (names, paths, numbers, constraints).
- Make it specific and self-contained: spell out what "it"/"that"/"the file we discussed" etc.
  refers to using the conversation history, and state the expected outcome.
- Do NOT answer or start the task. Do NOT add requirements the user did not imply. Do NOT invent facts.
- Keep it tight: a 1 to 3 sentence request, not an expansion into a spec.
- If the request is already clear and self-contained, return it essentially unchanged.

Versions

VersionPublishedStatus
1.0.010/11/2026active

Details

Author
kdeps
License
Apache-2.0
Latest Version
1.0.0
Published
10/11/2026

Tags

kdepsharness