/segmently-cli-content-plan-guide

Content Plan CLI Backlog

These are candidate CLI improvements for making Content Plan operations easier. Use this file only when planning product or CLI work.

Discovery

• content-plan designs discover <platformId> --project <projectId> --author <authorId>: return the platform's CLI-backed design profile sets, current profile IDs, all variants, reference counts, missing requested keys, and Design tab URLs. This should be the source-of-truth discovery surface for design-system-from-reference workflows.

Design Profile Operations

• content-plan designs references apply <platformId> ... --images ...: upload images, create visual reference rows, dry-run, and create/update a profile variant in one workflow.
• content-plan designs adapt <platformId> ... --source <file>: take one creative/design-system source plus a CLI-discovered target profile set and emit per-profile design manifests. The command should not depend on any skill-local hardcoded format table.
• content-plan designs apply-all <platformId> ... --file <package.json>: apply a package of per-profile manifests with a single dry-run summary and a single apply step after review.
• content-plan designs apply ... --no-set-current: explicit negative override for transfer manifests exported from current profiles. Today operators must edit profile.setCurrent=false in the manifest before applying a new variant.

Topics And Social Planning

• Implemented: content-plan authors get|apply|set-current|onboard, content-plan platforms list|get|apply, content-plan platforms voice get|apply, content-plan strategies get|apply|preflight|create, content-plan strategies draft create|patch|approve|regenerate, content-plan strategies sprints generate-topics, content-plan cadence get|apply, content-plan publications list|get|apply|patch|delete, content-plan posts create|schedule-batch, and content-plan bootstrap validate|plan|apply|verify cover the deterministic empty-project and reviewer-demo bootstrap path.
• Implemented in CLI 1.0.0: content-plan doctor reports the shared eight-step readiness chain and exact next commands; bootstrap validates by default and resumes from a manifest-hash checkpoint; task launches expose non-zero waited failures or taskState: "pending" plus followUp when not waited.
• Implemented: content-plan topics drafts list and content-plan topics accept <generationId> cover the generated-topic review flow. Remaining gap: bulk upsert a list of arbitrary topic manifests with create/update/delete operations and a single dry-run diff.
• Implemented: content-plan topics generate launches backlog topic generation from manual text and/or URL source context. By default URLs are passed as source metadata without scraping; --research-mode run_fresh|hybrid opts into fresh research using the same URLs as constrained sources.
• Implemented: content-plan topics resources import-url <topicId> stores an HTTPS URL as a link resource with optional title/content/reliability fields. Remaining gap: fetch and summarize remote page content into the resource.
• Implemented in CLI 1.0.0: client-side JSON Schema validation for bootstrap, topic apply, aspect apply, research phase patch, posts generate --file, and design-profile apply. These are the current JSON Schema manifests. Remaining schema gap: add the same client-side validator when new manifest-backed Content Plan surfaces are introduced.

Post And Asset Editing

• Implemented helpers that used to be gaps: content-plan posts accept <generationId>, content-plan posts refine <postId>, content-plan posts assets brief get <postId>, content-plan posts assets preview <postId>, content-plan posts assets slots regenerate-prompt <postId> <slotKey>, content-plan posts assets slots batch-generate, and content-plan posts assets slots experiments reject|delete.
• Implemented: content-plan posts drafts accept <draftPostId> and content-plan posts drafts reject <draftPostId> apply or reject one reviewed generated draft into canonical Content Plan storage from the CLI. Generation-level content-plan posts accept <generationId> also exists.
• content-plan posts regenerate-text / content-plan posts regenerate-post <postId>: implement true full writer rerun for an existing post when only the post ID is known. The current posts assets regenerate-text command bridges to the backend /regenerate-post endpoint, but that endpoint returns a clear conflict when stored generationPlanContext is missing. The final naming should separate copy-only refine from full writer+brief+asset reruns.
• content-plan posts assets slots regenerate-prompt currently regenerates a slot promptTemplate through project-scoped AI keys using the post brief and current slot as context. A deeper future version can call the exact storyboarding/card-visual pipeline stage for one carousel slide.
• Remaining schema candidates: post asset planning, brief regeneration, slot prompt patches, slot generation requests, and batch slot generation.