Research a topic with Hizal by checking existing context first, reading the relevant chunks, filling gaps from the repo or web, and writing back a focused summary.
Use this skill when the user wants research, discovery, or background gathering tied to Hizal.
Use it for requests like:
Start a session at the top of any research task — see hizal-onboard. End it with end_session when done.
Expect a Hizal MCP server to be configured with:
Authorization: Bearer <api-key>Choose the target project_id explicitly. If the project is unclear, call list_projects first.
list_projects when needed.search_context(query="<topic>", project_id="<project_id>", limit=5)read_contextwrite_knowledge for factual findings worth sharing with the team (PROJECT scope)write_memory for personal observations, failed approaches, and interpretive notes (AGENT scope)write_chunk(type="RESEARCH") for raw research notes that should be reviewed at end_session| What you found | Tool | Why |
|---|---|---|
| Architecture, patterns, factual discoveries | write_knowledge | Shared team context |
| A failed approach and why it failed | write_memory | Personal lesson, not team knowledge |
| Raw notes you want to review later | write_chunk(type="RESEARCH") | Ephemeral, surfaced at end_session |
| A decision with rationale | write_chunk(type="DECISION") | Long-lived, preserve carefully |
Never use write_convention, write_identity, or store_principle during research. Research findings are on-demand context — they should not flood every agent's context window.
Before using write_convention, ask:
If any answer is no, use write_knowledge or write_memory instead.
write_context.