DeepWiki: OSS repository architecture and topic map
Exa: breadth discovery and code/doc context
Perplexity: recent web-grounded synthesis and contradiction checks
WebSearch / WebFetch: standard web fallback or direct page retrieval
local repo tools: first-class evidence when the question touches local code
Output Contract
Always include:
Executive Summary
Evidence Map
What We Know
What We Infer
Contradictions and Unknowns
Confidence
Actionable Next Steps
Query Log
Sources
Citation Rules
Every material factual claim must have an inline citation.
Prefer direct URLs to primary sources.
If only secondary evidence exists, say so explicitly.
Keep facts and inference separate.
Failure Rules
If a tool is unavailable, switch to another tool with the same role.
If evidence is thin, say insufficient evidence.
If there is no primary source, lower confidence and state the limitation.
If evidence conflicts, preserve both sides with citations.
Report Template
Use this structure:
## Executive Summary
- Key takeaway with citation
## Evidence Map
- Primary: why this source is authoritative
- Supporting: what it adds
- Counter: what challenges the main claim
## What We Know
- Fact with citation
## What We Infer
- Inference linked back to evidence
## Contradictions and Unknowns
- Open disagreement or missing source
## Confidence
- high | medium | low
## Actionable Next Steps
- Specific follow-up
## Query Log
| Tool | Query/Prompt | Filters | Why Used | Outcome |
| --- | --- | --- | --- | --- |
## Sources
- [Title](https://example.com) - Publisher, date
Additional Guidance
For the detailed lane matrix and DeepWiki workflow, see reference.md.
For concrete example prompts and output shapes, see examples.md.
1---2name: source-grounded-research3description: Cited research briefs4---56# Source-Grounded Research78## Purpose910Use this skill for research-only tasks that should end in a cited brief rather11than code changes.1213This skill is for:1415- web research16- library and framework documentation lookup17- open-source repository investigation18- comparison and decision support19- recent changes and release-note checks2021Do not use this skill for:2223- implementation work24- code editing25- speculative answers without sources2627## Quick Start28291. Restate the question and classify it into one primary lane.302. Start from the lane's highest-authority source.313. Corroborate major claims with a second source.324. Check for contradictions or recent changes.335. Produce a cited report with a query log and sources section.3435## Research Lanes3637### `library-framework`3839- Start with official docs via `Context7` when available.40- Use `get_code_context_exa` for examples and practical usage.41- Use `Perplexity` or `WebSearch` only for contradiction checks or recent42 changes.4344### `oss-repository`4546- Start with `DeepWiki` when available.47- Read repo structure before asking detailed questions.48- Corroborate key claims with GitHub docs, repo pages, Exa, or official docs.4950### `recent-changes`5152- Start with `Perplexity Search` or standard `WebSearch`.53- Verify final claims with official release notes, changelogs, or docs.5455### `local-plus-external`5657- Start with local repo evidence tools first.58- Bridge to external docs only after the local behavior is understood.5960### `comparison-decision`6162- Start with breadth discovery.63- Run an authority pass.64- Run a contradiction pass before recommending.6566## Tool Priority6768Prefer tools by role, not habit:6970- `Sequential Thinking`: scope, sub-questions, assumptions71- `Context7`: official library/framework docs72- `DeepWiki`: OSS repository architecture and topic map73- `Exa`: breadth discovery and code/doc context74- `Perplexity`: recent web-grounded synthesis and contradiction checks75- `WebSearch` / `WebFetch`: standard web fallback or direct page retrieval76- local repo tools: first-class evidence when the question touches local code7778## Output Contract7980Always include:81821. Executive Summary832. Evidence Map843. What We Know854. What We Infer865. Contradictions and Unknowns876. Confidence887. Actionable Next Steps898. Query Log909. Sources9192## Citation Rules9394- Every material factual claim must have an inline citation.95- Prefer direct URLs to primary sources.96- If only secondary evidence exists, say so explicitly.97- Keep facts and inference separate.9899## Failure Rules100101- If a tool is unavailable, switch to another tool with the same role.102- If evidence is thin, say `insufficient evidence`.103- If there is no primary source, lower confidence and state the limitation.104- If evidence conflicts, preserve both sides with citations.105106## Report Template107108Use this structure:109110```markdown111## Executive Summary112- Key takeaway with citation113114## Evidence Map115- Primary: why this source is authoritative116- Supporting: what it adds117- Counter: what challenges the main claim118119## What We Know120- Fact with citation121122## What We Infer123- Inference linked back to evidence124125## Contradictions and Unknowns126- Open disagreement or missing source127128## Confidence129- high | medium | low130131## Actionable Next Steps132- Specific follow-up133134## Query Log135| Tool | Query/Prompt | Filters | Why Used | Outcome |136| --- | --- | --- | --- | --- |137138## Sources139- [Title](https://example.com) - Publisher, date140```141142## Additional Guidance143144- For the detailed lane matrix and DeepWiki workflow, see [reference.md](reference.md).145- For concrete example prompts and output shapes, see [examples.md](examples.md).
Run npx skillmds@latest add laststance/source-grounded-research in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.
Cited research briefs It is listed under Research & Search on SkillMD.
This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.
This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.
Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.
laststance (@laststance) published this skill. Their other Agent Skills are listed on their SkillMD profile.