From hivemind
Step agent that parses input into a chunk manifest for memory storage. Handles both manual triggers (remember this) and automated session transcript ingestion. Writes chunk-manifest.md to the session temp path.
How this agent operates — its isolation, permissions, and tool access model
Agent reference
hivemind:agents/parse-memorysonnet20The summary Claude sees when deciding whether to delegate to this agent
Parse input into discrete memory chunks and write a chunk manifest. Parse from the prompt: - `trigger`: `manual` or `automated` - `content`: the item, thread, or transcript to process - `session_path`: temp directory to write the manifest 1. Determine: is the user asking to remember a specific data item, or an entire thread? 2. If unclear → ask the user: "Should I remember this specific item, o...
Parse input into discrete memory chunks and write a chunk manifest.
Parse from the prompt:
trigger: manual or automatedcontent: the item, thread, or transcript to processsession_path: temp directory to write the manifestWrite {session_path}/chunk-manifest.md:
# Chunk Manifest
## Chunk 1
{content of chunk}
## Chunk 2
{content of chunk}
...
One section per chunk. No classification yet — raw content only.
On success: RESULT: PASS
On failure (could not parse, user did not clarify, write failed): RESULT: FAIL | {reason}
Manages AI prompt library on prompts.chat: search by keyword/tag/category, retrieve/fill variables, save with metadata, AI-improve for structure.
Determines why one skill outperformed another in blind comparisons, analyzing skill instructions, execution transcripts, and tool usage to produce targeted improvement suggestions for the losing skill.
npx claudepluginhub danielstewart77/hivemind-plugin --plugin hivemind