# Knowledge Graph

> Persist research graph of Question/Claim/Evidence/Source relations. Use to keep a complex research task's findings structured and reusable across sessions.

- Skill: `samraddhashrivastavatech/knowledge-graph` (Agent Skill)
- Install (CLI): `npx skillmds@latest add samraddhashrivastavatech/knowledge-graph`
- Raw SKILL.md: https://api.skillmd.com/api/skills/samraddhashrivastavatech/knowledge-graph/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Research & Search
- Author: SAMRADDHASHRIVASTAVATECH (https://skillmd.com/u/samraddhashrivastavatech)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/samraddhashrivastavatech/knowledge-graph

---


# Knowledge graph

Persist the research structure — questions, claims, evidence, sources and their
relations — so multi-step research stays coherent.

## When to use

- A research task with many claims and sources that must not turn into a list.
- Handing off a research state to be reused or audited.
- Warming a `deep-research` objective with existing structured findings.

## Tools used

- `research_graph` — retrieve the node/edge graph of a research task.
- `research_status` — current state of a research task.

## Workflow

1. As evidence is extracted, keep claim → evidence → source → location tuples.
2. `research_graph` to persist/retrieve the structured relations (Q/C/E/S).
3. `research_status` to keep track of progress on long tasks.
4. Walk the graph for gaps (claims without evidence) before finalizing.

## Input schema

```json
{ "task_id": "str" }
```

## Output schema

```json
{ "task_id": "str", "nodes": "list[object]", "edges": "list[object]" }
```

## Security

The graph stores claims and sources; it stores no record of who/what was
verified. Keep it task-scoped; never fold private or sensitive material into a
shared graph.

## Related skills

`evidence-extraction`, `deep-research`, `claim-verification`,
`literature-review`, `document-research`