feat: add content creation skill for composing primitives#6
Open
recoupableorg wants to merge 13 commits intomainfrom
Open
feat: add content creation skill for composing primitives#6recoupableorg wants to merge 13 commits intomainfrom
recoupableorg wants to merge 13 commits intomainfrom
Conversation
Teaches LLM agents how to use the modular content creation primitives (image, video, text, audio, render, upscale) via CLI. Includes step-by-step and quick full-pipeline workflows, template selection guidance, and iteration strategies. Made-with: Cursor
📝 WalkthroughWalkthroughAdded Changes
Estimated code review effort🎯 2 (Simple) | ⏱️ ~8 minutes Poem
🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
- Pushy description with comprehensive trigger phrases - Creative decision-making guidance (template selection, song choice, quality evaluation) instead of just procedural steps - Explains WHY (e.g. start with audio because mood informs template) - Edge case handling (instrumentals, missing face guide, song URLs) - Iteration strategies showing cost savings of modular approach Made-with: Cursor
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Nitpick comments (1)
content-creation/SKILL.md (1)
44-51: Consider converting prose to bullet points.The "Choosing a song clip" section is prose-heavy. Per coding guidelines, bullet points and numbered lists are preferred.
♻️ More concise structure with bullets
### Choosing a song clip -Start with audio before image because the song's mood should influence your template choice. If the artist has multiple songs, pick one that matches the content goal: +Start with audio before image — the song's mood influences your template choice. + +**When to target a specific song:** - Promoting a new release? Use `--song <slug>` to target it - General content? Let the pipeline pick randomly - Have a specific audio file? Pass the URL: `--song https://...` -After audio selection completes, check `clipMood` and `clipLyrics` in the output. If the mood doesn't match the template you planned, switch templates before generating the image. +**After audio selection:** Check `clipMood` and `clipLyrics`. If the mood doesn't match your planned template, switch templates before generating the image.As per coding guidelines: "Prefer bullet points and numbered lists over prose in SKILL.md files"
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@content-creation/SKILL.md` around lines 44 - 51, Convert the prose under the "Choosing a song clip" section into concise bullet points and/or a short numbered list: keep the header "Choosing a song clip", then list steps such as "Start with audio before image", selection options (use `--song <slug>` for a new release, let pipeline pick for general content, pass URL `--song https://...` for a specific file), and a final bullet instructing to check `clipMood` and `clipLyrics` and switch templates if mood mismatches; preserve the existing command examples and the `clipMood`/`clipLyrics` identifiers so readers can find the relevant output fields.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@content-creation/SKILL.md`:
- Line 29: The SKILL.md text refers to a non-existent resource
"context/artist.md"; update SKILL.md (content-creation/SKILL.md) to either point
to an existing file or add the missing artist context: create a new
context/artist.md with the artist’s personality, aesthetic direction, mood,
colors, and sacred rules, or change the sentence to reference the correct
existing resource (e.g., an actual path or a shared README) or instruct readers
to consult an inline "Artist Context" section within SKILL.md; ensure the
reference string "context/artist.md" is removed or replaced so it matches the
repository’s actual structure.
- Around line 73-115: Replace all XML-style angle-bracket placeholders in
SKILL.md (e.g., occurrences in the command examples like "recoup content audio
--artist <id> --json", "recoup content image --artist <id> --template <name>
--json", "recoup content video --image <imageUrl> --json", "recoup content
render --video <videoUrl> --audio <songUrl> --start <startSeconds> --duration 15
--text \"<content>\" --json", and prose mentions of <id>, <runId>, <slug>,
<url>, etc.) with one of the approved formats (ALL_CAPS tokens like ARTIST_ID,
RUN_ID, SONG_SLUG, IMAGE_URL; or {id}/{runId}/{slug}/{url}; or
{{id}}/{{runId}}/{{slug}}/{{url}}) consistently throughout the file, ensuring
each placeholder type is replaced in both prose and code blocks (examples and
quick/lipsync snippets) so no angle brackets remain.
---
Nitpick comments:
In `@content-creation/SKILL.md`:
- Around line 44-51: Convert the prose under the "Choosing a song clip" section
into concise bullet points and/or a short numbered list: keep the header
"Choosing a song clip", then list steps such as "Start with audio before image",
selection options (use `--song <slug>` for a new release, let pipeline pick for
general content, pass URL `--song https://...` for a specific file), and a final
bullet instructing to check `clipMood` and `clipLyrics` and switch templates if
mood mismatches; preserve the existing command examples and the
`clipMood`/`clipLyrics` identifiers so readers can find the relevant output
fields.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
Commands renamed to verb-qualifier pattern: generate-image, generate-video, generate-caption, transcribe-audio. Made-with: Cursor
- Removed music-specific language and param names - Added model selection docs and edit operations reference - Updated all workflow examples with new CLI flags - Added analyze-video to primitives table Made-with: Cursor
Teaches the agent to watch its own generated videos using analyze-video after every generation step, evaluate quality, and iterate based on the feedback. Made-with: Cursor
Two distinct evaluation passes: QA (artifacts, glitches, motion) after generation, and taste (hook, aesthetic, platform readiness) after the final edit. Scoring rubric guides iteration decisions. Made-with: Cursor
Gives guidance on what to look for (QA, taste, platform readiness) and how to act on feedback, but lets the agent write its own prompts. Made-with: Cursor
…lows Analyze at every creative checkpoint: after each clip, each edit pass, multi-clip assembly, audio sync, text overlay, and version comparison. Fix issues early before building on top of them. Made-with: Cursor
Address CodeRabbit review: skills repo guidelines prohibit XML-style
angle brackets in SKILL.md files. All placeholders now use {curly}
brace syntax.
Made-with: Cursor
Image is optional — video can be generated from just a prompt. Made-with: Cursor
…ples - New Video Generation Modes section with mode table and when-to-use guide - Updated model selection table with all auto-selected models - Added extend and first-last workflow examples - Updated lipsync to use --mode lipsync instead of --lipsync flag - Added extend and first-last to iteration and acting-on-feedback sections Made-with: Cursor
All commands now use short nouns: image, video, caption, transcribe, analyze, edit, upscale. Made-with: Cursor
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (2)
content-creation/SKILL.md (2)
30-30: Use one command naming convention consistently (noun endpoint names).This file mixes
generate-video/analyze-videolabels with noun-based commands (recoup content video,recoup content analyze). Standardizing terminology will reduce operator confusion and better match the PR’s endpoint naming direction.As per coding guidelines: "Prefer bullet points and numbered lists over prose in SKILL.md files; place critical instructions at the top of the file."
Also applies to: 125-133, 151-151, 202-207
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@content-creation/SKILL.md` at line 30, The SKILL.md currently mixes verb-based commands like `generate-video` and `analyze-video` with noun-style CLI usages such as `recoup content video` and `recoup content analyze`; update the file to use one consistent naming convention (prefer the noun endpoint names: `recoup content video`, `recoup content analyze`) across all occurrences (including the sections referenced around lines 125-133, 151, and 202-207), convert the prose instructions into bullet points/numbered lists, and move the critical usage note about `--mode` (or input-inferred mode) to the top of the document so operators see it first.
24-24: Document non-success terminal states for async polling.Current guidance only says to poll until
COMPLETED. Add explicit handling forFAILED/CANCELLED(and what to do next) to avoid infinite polling loops in agent workflows.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@content-creation/SKILL.md` at line 24, Update the async polling guidance around the recoup tasks status CLI (the line mentioning "Poll with `recoup tasks status --run {runId} --json` until `status` `COMPLETED`") to enumerate non-success terminal states and next actions: explicitly document that `FAILED` and `CANCELLED` are terminal, instruct callers to stop polling on those values, surface error/exit messages from the task output for `FAILED`, suggest retry or investigate steps and how to cancel/cleanup for `CANCELLED`, and show the recommended behavior for automation (e.g., fail the workflow, raise an error, or trigger retry/backoff) so agent loops cannot poll forever.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@content-creation/SKILL.md`:
- Around line 81-83: The example command uses --audio while the operations table
documents --mux-audio, causing inconsistency; pick the actual CLI flag
(preferably --mux-audio if the CLI performs muxing) and update both the example
command "recoup content edit --video {videoUrl} --audio {audioUrl} ...
--overlay-text" and the operations table entries to the same flag (either
replace --audio with --mux-audio everywhere or vice versa) so both examples and
docs match the real CLI behavior.
---
Nitpick comments:
In `@content-creation/SKILL.md`:
- Line 30: The SKILL.md currently mixes verb-based commands like
`generate-video` and `analyze-video` with noun-style CLI usages such as `recoup
content video` and `recoup content analyze`; update the file to use one
consistent naming convention (prefer the noun endpoint names: `recoup content
video`, `recoup content analyze`) across all occurrences (including the sections
referenced around lines 125-133, 151, and 202-207), convert the prose
instructions into bullet points/numbered lists, and move the critical usage note
about `--mode` (or input-inferred mode) to the top of the document so operators
see it first.
- Line 24: Update the async polling guidance around the recoup tasks status CLI
(the line mentioning "Poll with `recoup tasks status --run {runId} --json` until
`status` `COMPLETED`") to enumerate non-success terminal states and next
actions: explicitly document that `FAILED` and `CANCELLED` are terminal,
instruct callers to stop polling on those values, surface error/exit messages
from the task output for `FAILED`, suggest retry or investigate steps and how to
cancel/cleanup for `CANCELLED`, and show the recommended behavior for automation
(e.g., fail the workflow, raise an error, or trigger retry/backoff) so agent
loops cannot poll forever.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
content-creation/SKILL.md
Outdated
Comment on lines
+81
to
+83
| recoup content edit --video {videoUrl} --audio {audioUrl} \ | ||
| --template artist-caption-bedroom \ | ||
| --overlay-text "{caption}" --json |
There was a problem hiding this comment.
Fix edit audio flag inconsistency (--audio vs --mux-audio).
The workflow example uses --audio, but the operations table defines muxing as --mux-audio. One of these is wrong and will cause failed commands if copied as-is. Please align both sections to the actual CLI flag.
Also applies to: 144-145
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@content-creation/SKILL.md` around lines 81 - 83, The example command uses
--audio while the operations table documents --mux-audio, causing inconsistency;
pick the actual CLI flag (preferably --mux-audio if the CLI performs muxing) and
update both the example command "recoup content edit --video {videoUrl} --audio
{audioUrl} ... --overlay-text" and the operations table entries to the same flag
(either replace --audio with --mux-audio everywhere or vice versa) so both
examples and docs match the real CLI behavior.
…priority - Lead with malleable-first philosophy (templates optional) - Add analyze primitive to table and workflow - Rename workflows: "Without a template" / "With a template" - Update edit command to use --url (supports images too) - Add template override priority section - Update frontmatter description Made-with: Cursor
There was a problem hiding this comment.
♻️ Duplicate comments (1)
content-creation/SKILL.md (1)
83-85:⚠️ Potential issue | 🟠 MajorAlign
editaudio flag usage (--audiovs--mux-audio).The workflow example uses
--audio, but the operations table defines audio muxing as--mux-audio. Please standardize on the actual CLI flag in both places to avoid broken copy/paste commands.Also applies to: 151-151
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@content-creation/SKILL.md` around lines 83 - 85, The example invocation of the CLI uses the wrong audio flag: update the `recoup content edit` example to use the canonical `--mux-audio` flag (instead of `--audio`) so it matches the operations table; specifically change the flag in the sample command that includes `--template artist-caption-bedroom --overlay-text "{caption}" --json` to `--mux-audio {audioUrl}` so copy/paste works correctly.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Duplicate comments:
In `@content-creation/SKILL.md`:
- Around line 83-85: The example invocation of the CLI uses the wrong audio
flag: update the `recoup content edit` example to use the canonical
`--mux-audio` flag (instead of `--audio`) so it matches the operations table;
specifically change the flag in the sample command that includes `--template
artist-caption-bedroom --overlay-text "{caption}" --json` to `--mux-audio
{audioUrl}` so copy/paste works correctly.
sidneyswift
added a commit
that referenced
this pull request
Apr 4, 2026
AI-powered content pipeline — image generation, video generation, captions, editing, upscaling, and iterative QA via analyze. Supports templates and freeform prompts. Made-with: Cursor
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Rewrite content-creation skill for the new generic content primitive system.
Summary by CodeRabbit