How do I install the Documentation Research skill?

Run npx skillmds@latest add tomevault-io/documentation-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.

Is Documentation Research safe to use?

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.

Which AI agents work with Documentation Research?

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.

Is Documentation Research free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Documentation Research?

tomevault-io (@tomevault-io) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Documentation Research by tomevault-io

Name: Documentation Research
Author: tomevault-io

Documentation Research

Fetch and cite current upstream documentation for every external dependency in scope. ALWAYS use before writing code against an external SDK/framework/API/CLI, during forge Phase 2, or whenever training data may be stale. Produces ISO-8601-stamped source files under evidence/documentation-research/ and a SUMMARY.md citing 3-5 verified facts per source. Refuses memory-only references — every external fact must cite a sources/ file. Use when this capability is needed.

tomevault-io 1189360 2 files · 2.7 KB Updated Sep 17, 2026

File contents

documentation-research

Phase 2 of /shannon:forge. Grounds every external dependency claim in fetched documentation, not training-data recall.

Behavior contract

Produces evidence/documentation-research/<run-id>/ containing:

sources/ — raw markdown of each fetched documentation page, named <source>-<ISO-8601-timestamp>.md.
SUMMARY.md — citing 3-5 verified facts per source, each linking to the local sources/ filename.
README.md + INDEX.md.

Methodology

For each external dependency identified by codebase-analysis:

Prefer context7 MCP for library docs (current, version-aware).
Fall back to WebFetch against the canonical docs URL.
Save the raw markdown to sources/<lib>-<timestamp>.md.
In SUMMARY.md, write 3-5 facts with [source: sources/<file>] inline citations.

When to use

Phase 2 of /shannon:forge.
Before any SDK / API integration.
When training data may be stale (recently released libraries, breaking changes in current major).

When NOT to use

Pure internal-codebase work with no external deps.
Already-fetched docs < 7 days old (cite the existing source file instead).

Iron rules

Refuse memory-only references. Every external-fact claim must cite a sources/ file.
ISO-8601 timestamps mandatory. Stale docs are surfaced, not silently used.
No invented URLs. WebFetch only against URLs the user provided or context7 returned.

Source: krzemienski/shannon — distributed by TomeVault.

---name: documentation-researchdescription: Fetch and cite current upstream documentation for every external dependency in scope. ALWAYS use before writing code against an external SDK/framework/API/CLI, during forge Phase 2, or whenever training data may be stale. Produces ISO-8601-stamped source files under evidence/documentation-research/ and a SUMMARY.md citing 3-5 verified facts per source. Refuses memory-only references — every external fact must cite a sources/ file. Use when this capability is needed.---# documentation-researchPhase 2 of `/shannon:forge`. Grounds every external dependency claim in fetched documentation, not training-data recall.## Behavior contractProduces `evidence/documentation-research/<run-id>/` containing:1. `sources/` — raw markdown of each fetched documentation page, named `<source>-<ISO-8601-timestamp>.md`.2. `SUMMARY.md` — citing 3-5 verified facts per source, each linking to the local `sources/` filename.3. `README.md` + `INDEX.md`.## MethodologyFor each external dependency identified by `codebase-analysis`:1. Prefer `context7` MCP for library docs (current, version-aware).2. Fall back to `WebFetch` against the canonical docs URL.3. Save the raw markdown to `sources/<lib>-<timestamp>.md`.4. In `SUMMARY.md`, write 3-5 facts with `[source: sources/<file>]` inline citations.## When to use- Phase 2 of `/shannon:forge`.- Before any SDK / API integration.- When training data may be stale (recently released libraries, breaking changes in current major).## When NOT to use- Pure internal-codebase work with no external deps.- Already-fetched docs < 7 days old (cite the existing source file instead).## Iron rules- **Refuse memory-only references.** Every external-fact claim must cite a `sources/` file.- **ISO-8601 timestamps mandatory.** Stale docs are surfaced, not silently used.- **No invented URLs.** WebFetch only against URLs the user provided or context7 returned.---> Source: [krzemienski/shannon](https://github.com/krzemienski/shannon) — distributed by [TomeVault](https://tomevault.io).

Frequently asked questions