content-ops

Identifies content gaps and appends net-new suggestions to the content backlog. Operates in two modes: triggered automatically by write-content after a piece is committed (orchestrator mode), or invoked manually via /suggest-content (user mode). In orchestrator mode, scans the newly written article for concepts — people, technologies, events, terms — that are mentioned but lack dedicated content. In user mode, surveys the full content corpus against pillars to find the most significant gaps across all themes. Always deduplicates against the content index and existing backlog before appending. <example> Context: write-content has just committed "proof-of-work.md" and is invoking backlog-suggester to find follow-on content gaps. user: "triggered_by: content/articles/proof-of-work.md, mode: orchestrator" assistant: "I'll use backlog-suggester to scan the new article for unaddressed concepts and append any net-new gaps to the backlog." <commentary> Orchestrator mode — write-content passes the committed file path; the agent scans it for mentioned concepts (e.g. "Hal Finney" appears twice, no article exists) and appends high-priority suggestions without user interaction. </commentary> </example> <example> Context: The user has run /suggest-content and wants gap analysis focused on a specific theme. user: "articles related to bitcoin mining" assistant: "I'll use backlog-suggester to survey the corpus for mining-adjacent pillar gaps and surface the top candidates with rationale." <commentary> User mode — the payload narrows scope to mining topics; the agent scores candidates by prerequisite coverage and theme balance, runs a timeliness check, then asks the user which suggestions to write to the backlog. </commentary> </example>

Handles bidirectional linking between articles and glossary entries. Uses content-index.json for discovery, one bounded LLM ranking pass, then edits only the selected files. Token use is capped via linking_max_candidates and linking_max_links. <example> Add bidirectional links between src/content/articles/en/liquid-staking.md and related glossary entries and articles using content-index.json. </example>

Verifies factual claims and researches topics for content creation. Uses a file-based research cache to avoid re-verifying known facts. Use this agent when writing, reviewing, or fact-checking content. <example> Research and verify facts for an article on "proof of stake": confirm consensus mechanism details, find authoritative sources, check current validator counts. </example> <example> Fact-check the claim: "Ethereum processes 15 transactions per second on mainnet" </example>

Writes content files for any configured content type. Receives a content type name, topic, research findings, output path, and config values from the write-content orchestrator. Follows the type-specific structure guide loaded via the content-style skill. Use this agent when you need to draft content files — do not use it for research, linking, or review. <example> Draft a new article on "liquid staking" to src/content/articles/en/liquid-staking.md using the provided research findings. </example> <example> Write a glossary entry for "validator" to src/content/glossary/en/validator.md using the provided research findings. </example> <example> Write a case-study on "Project X migration" to src/content/case-studies/en/project-x-migration.md using the provided research findings. </example>

Scans a newly written article for technical terms that lack glossary entries and creates minimal stubs for them. Use after the draft-writer agent, before the content-linker agent. Do not use for writing full glossary entries — use draft-writer for that. <example> Scan src/content/articles/en/liquid-staking.md for technical terms missing from src/content/glossary/en/ and create stubs for any gaps. </example>

Image style reference — visual guidelines, prompt templates, placement rules, and file naming conventions for image generation. Reads project-specific values from plugin config.

Auto-injects current article and glossary inventory for content creation and linking tasks. Reads paths from plugin config.

Loads and applies the project's configured style guide. Hard stops if config or any required piece is missing. All style rules come from the guidelines file set during /init style.

Fact-check content by verifying every claim against trusted sources.

content-ops setup wizard. Run /init to see status, or /init [round] to run a setup step. Rounds: project, content-types, style, strategy, infra, images, link-building.

2 hooks across 1 event