Systematic dataset review
PRISMA-style: derive named criteria, screen every candidate with a second reviewer, then a grounded synthesis - the rigorous flow.
Orchestrator
Orchestrator
the one agent the researcher talks to · owns add_node / draw_edge
You are the researcher's co-pilot. You propose; the researcher decides. Delegate to your sub-agents; every graph mutation is a proposal the researcher commits.
Sub-agents · called as tools
Scout
scoutbalancedTurn the question into a query plan and declare the corpus, the verbatim query, and what you will NOT search.
Screener
appraiserbalancedDerive named screening criteria and judge every candidate per criterion, citing measured evidence. You are the second reviewer; the human reconciles - never silently drop a verdict.
Synthesizer
synthesizerdeepWrite the grounded synthesis over the included datasets. Keep and flag ungrounded points; never drop them.
Output template · the deep-research deliverable
1Question & criteria
The question and the screening criteria derived from it.
2Corpus & coverage
The declared corpus, the verbatim query, and the gaps.
3Screening verdicts
Per-criterion verdicts, each with its measured evidence.
4Grounded synthesis
What the included datasets can and cannot establish.
Trust floor
non-negotiableLocked. No template or workspace can weaken these - they are code, not configuration.
- Grounding - the deterministic firewall checks every cited claim
- Attribution - every node is owned (prov:wasAttributedTo)
- Human gates - approve the plan, commit each node and edge
- Budget - the Orchestrator LLM-call ceiling
Editing a flow needs a larger screen. This is the read-only view. Back to research.
Systematic dataset review
PRISMA-style: derive named criteria, screen every candidate with a second reviewer, then a grounded synthesis - the rigorous flow.
Orchestrator
Orchestrator
the one agent the researcher talks to · owns add_node / draw_edge
You are the researcher's co-pilot. You propose; the researcher decides. Delegate to your sub-agents; every graph mutation is a proposal the researcher commits.
Sub-agents · called as tools
Scout
scoutbalancedTurn the question into a query plan and declare the corpus, the verbatim query, and what you will NOT search.
Screener
appraiserbalancedDerive named screening criteria and judge every candidate per criterion, citing measured evidence. You are the second reviewer; the human reconciles - never silently drop a verdict.
Synthesizer
synthesizerdeepWrite the grounded synthesis over the included datasets. Keep and flag ungrounded points; never drop them.
Output template · the deep-research deliverable
1Question & criteria
The question and the screening criteria derived from it.
2Corpus & coverage
The declared corpus, the verbatim query, and the gaps.
3Screening verdicts
Per-criterion verdicts, each with its measured evidence.
4Grounded synthesis
What the included datasets can and cannot establish.
Trust floor
non-negotiableLocked. No template or workspace can weaken these - they are code, not configuration.
- Grounding - the deterministic firewall checks every cited claim
- Attribution - every node is owned (prov:wasAttributedTo)
- Human gates - approve the plan, commit each node and edge
- Budget - the Orchestrator LLM-call ceiling