tome

Multi-source research plugin for code archaeology, community discourse, academic literature, and TRIZ cross-domain analysis.

Overview

Tome orchestrates research across four channels: GitHub code search, community discourse (HN, Lobsters, Reddit), academic literature (arXiv, Semantic Scholar), and TRIZ analogical reasoning. It classifies domains and adapts search depth automatically.

Frontier detection

A research report ends with a verdict describing the search itself. Few findings can mean a thin field or a search that went wrong, and those two cases call for opposite responses. To tell them apart, each channel runs a positive control: a query whose answer is known to be in that channel’s index. A channel that cannot retrieve a document it should find is blind, and its silence carries no information about the field.

Rules are evaluated in order, and the first match wins:

Verdict	Meaning
`INCONCLUSIVE`	A channel errored, was rate-limited, or failed its control. The run is not evidence.
`THIN_FIELD_CANDIDATE`	Two or more channels searched cleanly and still returned few findings.
`CHANNEL_MISMATCH_SUSPECTED`	One channel holds most findings while a control-passing channel came back empty, which points to a vocabulary mismatch more often than to a thin field.
`COVERED`	The channels searched cleanly and returned enough findings that no sparsity question arises.

Only channels answering from an external index take part in the verdict. RETRIEVAL_CHANNELS in tome/models.py names them: academic, code, and discourse. The triz channel writes analogies rather than retrieving records, and its exclusion holds in two directions. Counting its output would let a session clear the sparsity threshold with text it wrote itself. Demanding a control from a channel that has no index would pin every session carrying triz to INCONCLUSIVE for a reason that says nothing about coverage.

A null result is reported with the floor it was measured against. For the rationale and the limits of the mechanism, see ADR-0020.

Installation

/plugin install tome@claude-night-market

Commands

Command	Description
`/tome:research`	Run multi-source research session
`/tome:dig`	Refine results interactively
`/tome:cite`	Generate formatted bibliography
`/tome:export`	Export findings for knowledge-intake

Skills

research: orchestrate a full research session
code-search: search GitHub implementations
discourse: scan community discussions
papers: search academic literature
triz: cross-domain analogical reasoning
ideate: diverse ideation methods with rotation
synthesize: merge and rank findings
dig: interactive refinement

Agents

code-searcher: GitHub code search
discourse-scanner: community discussion scanning
literature-reviewer: academic paper review
research: multi-source research orchestrator (delegates to Skill(tome:research))
triz-analyst: cross-domain analysis

Keyboard shortcuts