Raise LinkedIn article word-count floor to 2,200w#42
Merged
Conversation
The default linkedin-article skill contract targeted 700-1,000 words, which is wrong for this channel — per feedback the LinkedIn article floor is 2,200 words. Update the contract's word_count range and deterministic check, and align every stale downstream target that fed generation off the old number: - harness/skill-contracts/linkedin-article.yaml: word_count 700-1000 -> 2200-3000; deterministic word_count_range check raised to 2200-3000 - scripts/content/linkedin_article_writer.py: prompt target 900-1,200 -> 2,200-3,000 (floor); raise max_tokens 2000 -> 4500 so a 2,200w+ article is not truncated - scripts/content/_writer.py: engine linkedin format instruction 600-900 -> 2,200-3,000 (this format maps to the linkedin-article policy) - knowledge/channels/linkedin-articles.md: length spec 1,200-2,000 -> 2,200-3,000 across quick-ref, checklists, and worked examples - harness/skills/kai-repurpose/SKILL.md: LinkedIn article variant 700-1000 -> 2,200-3,000 - harness/skill-contracts/social-post.yaml: article cross-reference comment updated to 2200-3000 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_016dF1TwvB2u4jxSKhbyERr6
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What & why
The default
linkedin-articleskill contract targeted 700–1,000 words, which is wrong for this channel. Per Connor's LinkedIn writing feedback, the LinkedIn article floor is 2,200 words. The deployed writer on the agent box was still generating to the stale 700–1,000w (its own prompt said 900–1,200w), so drafts came out far shorter than the channel warrants.This PR raises the floor to 2,200 words and aligns every stale downstream target that fed generation off the old number, so the contract, the channel guide, and the writers all agree.
Changes
harness/skill-contracts/linkedin-article.yaml—word_count700-1000→2200-3000;word_count_rangedeterministic check raised to2200 <= word_count <= 3000. This is the source of truth the writer derives its floor target from.scripts/content/linkedin_article_writer.py(deployed agent-box writer) — prompt length900–1,200→2,200–3,000(floor);max_tokens2000→4500so a 2,200w+ article is not truncated mid-generation.scripts/content/_writer.py— enginelinkedinformat instruction600-900→2,200-3,000(this format maps to thelinkedin-articlepolicy, so the old number contradicted the contract in-prompt).knowledge/channels/linkedin-articles.md— length spec1,200-2,000→2,200-3,000across the quick-reference table, checklists, and worked examples.harness/skills/kai-repurpose/SKILL.md— LinkedIn article variant700-1000→2,200-3,000.harness/skill-contracts/social-post.yaml— article cross-reference comment updated to2200-3000(the LinkedIn organic-post range is unchanged).Verification
python scripts/quality_gates/golden_check.py→ 4/4 golden cases intact.python scripts/doctor.py→ harness intact (45/45 referenced files, instruction chain, gate scripts compile). Remaining warnings are pre-existing unconfigured-env warnings, unrelated to this change.2200.Note
The ceiling (3,000w) is not specified by the feedback — the feedback fixes only the 2,200 floor. I paired it with a 3,000 ceiling to keep a bounded range for the deterministic check and writer target (the prior contract and channel guide both used bounded ranges). Easy to widen if you'd prefer a higher or open-ended ceiling.
🤖 Generated with Claude Code
https://claude.ai/code/session_016dF1TwvB2u4jxSKhbyERr6
Generated by Claude Code