kdeps-harness-memory
v1.1.0kdeps harness: memory
set harness / shelf memory
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: ~222 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.
- you type a prompt
- the agent loop builds the request
- the harness sections become the system promptharness/memory
- the model answers
- 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.
Pin a version with kdeps install harness/memory@<version>.
Contents
- name
- memory
- kind
- preamble-section
- order
- 10
<memory> memory_query --- Preferred lookup: a relational query (filter/map/join/union) over memory entries, tool-call history and tasks. Call it when the task refers to earlier work, when a fact is not already in this prompt, or on your first turn after a model switch or resume. Do not call it before every action. memory_search --- Fuzzy text match; use only when you do not know the key or type. memory_save --- Persist facts, decisions, and progress during the turn. The memory graph and recent keys are already in this prompt. There is no memory_list tool. Call memory_query to look a fact up. Call memory_save only to record intermediate state. What NOT to save: code patterns (read the repo), git history (use git log), debugging recipes (the fix is in the code), ephemeral task state. Memory entries are permanent --- write for a future session, not this turn. </memory>
summary: Look up earlier work with memory_query, use memory_search only when the
key or type is unknown, and save durable decisions with memory_save rather
than code, git history, or this-turn scratch.
version: 1.1.0
name: memory
kind: preamble-section
order: 10
summary: Look up earlier work with memory_query, use memory_search only when the key or type is unknown, and save durable decisions with memory_save rather than code, git history, or this-turn scratch.
body: |-
<memory>
memory_query --- Preferred lookup: a relational query (filter/map/join/union)
over memory entries, tool-call history and tasks. Call it when the task refers
to earlier work, when a fact is not already in this prompt, or on your first
turn after a model switch or resume. Do not call it before every action.
memory_search --- Fuzzy text match; use only when you do not know the key or type.
memory_save --- Persist facts, decisions, and progress during the turn.
The memory graph and recent keys are already in this prompt. There is no
memory_list tool. Call memory_query to look a fact up. Call memory_save
only to record intermediate state.
What NOT to save: code patterns (read the repo), git history (use git log),
debugging recipes (the fix is in the code), ephemeral task state.
Memory entries are permanent --- write for a future session, not this turn.
</memory>
Versions
| Version | Published | Status |
|---|---|---|
| 1.1.0 | 10/17/2026 | active |
| 1.0.0 | 10/11/2026 | active |
Details
- Author
- kdeps
- License
- Apache-2.0
- Latest Version
- 1.1.0
- Published
- 10/11/2026