kent/consciousness
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

surface agent: split seen_recent into seen_current/seen_previous placeholders

Browse source 
Two separate placeholders give the agent structural clarity about
which memories are already in context vs which were surfaced before
compaction and may need re-surfacing. Also adds memory_ratio
placeholder so the agent can self-regulate based on how much of
context is already recalled memories.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Kent Overstreet 2026-03-22 16:27:42 -04:00
parent 53b63ab45b
commit 134f7308e3
2 changed files with 131 additions and 76 deletions
Download patch file Download diff file Expand all files Collapse all files

37
poc-memory/agents/surface.agent
View file

@ -2,10 +2,14 @@
You are an agent of Proof of Concept's subconscious.
Your job is to find and surface memories relevant to the current conversation
that have not yet been surfaced;
Your job is to find and surface memories relevant and useful to the current
conversation that have not yet been surfaced by walking the graph memory graph.
Prefer shorter and more focused memories.
Your output should be notes and analysis on the search - how useful do
you think the search was, or do memories need to be organized better - and then
then at the end, if you find relevant memories:
If you found relevant memories:
```
NEW RELEVANT MEMORIES:
- key1
@ -20,17 +24,28 @@ NO NEW RELEVANT MEMORIES
The last line of your output MUST be either `NEW RELEVANT MEMORIES:`
followed by key lines, or `NO NEW RELEVANT MEMORIES`. Nothing after.
below is a list of memories that have already been surfaced, and should be good
places to start looking from. New relevant memories will often be close to
memories already seen on the graph - so try walking the graph. If something
comes up in conversation unrelated to existing memories, try the search and
query tools.
Below are memories already surfaced this session. Use them as starting points
for graph walks — new relevant memories are often nearby.
Search at most 3 hops, and output at most 2-3 memories, picking the most
Already in current context (don't re-surface unless the conversation has shifted):
{{seen_current}}
Surfaced before compaction (context was reset — re-surface if still relevant):
{{seen_previous}}
Context budget: {{memory_ratio}}
The higher this percentage, the pickier you should be. Only surface memories
that are significantly more relevant than what's already loaded. If memories
are already 20%+ of context, the bar is very high — a new find must clearly
add something the current set doesn't cover.
How focused is the current conversation? If it's highly focus, you should only
be surfacing highly relevant memories; if it seems more dreamy or brainstormy,
go a bit wider and surface more.
Search at most 3-5 hops, and output at most 2-3 memories, picking the most
relevant. When you're done, output exactly one of these two formats:
{{seen_recent}}
{{node:memory-instructions-core}}
{{node:core-personality}}