open/linkedin-studio
1
0
Fork 0
Code Issues Pull requests Releases Wiki Activity Actions

fix(linkedin-studio): S3 harden react — Step-6 buzzword gate + WebFetch-failure fallback (both paths)

Browse source 
S3 close-out (3 edits / 1 file · lint 81/0 · counts 29/19/26/6):
- react.md:131 Step 6: add no-buzzword checklist line (was missing; banned list = content rule #4; covers Step 6b via "Same as Step 6")
- react.md:60 Step 2: WebFetch-failure fallback (paywall/JS/YouTube-no-transcript -> ask user to paste or pick another; never fabricate)
- react.md:194 Step 2b: same fallback pointer for the multi-URL comparison path

Axis-c (buzzword) + axis-d (graceful degradation) closed on both paths; axes a/b PASS. Shared scorecard already clean post-S2. Local-only per push policy.
This commit is contained in:
Kjell Tore Guttormsen 2026-06-18 13:44:00 +02:00
commit b8c2155003
2 changed files with 79 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

5
commands/react.md
View file

@ -57,6 +57,8 @@ Use WebFetch to extract the content from the URL. Ask WebFetch to extract:
- Data points or statistics mentioned
- The "so what" — why this matters
**If WebFetch can't extract usable content** (paywall, JS-heavy page, or a YouTube video with no accessible transcript — all accepted inputs WebFetch may not resolve): say so plainly, then ask the user to paste the key points (or the transcript) or choose another URL. Never fabricate a reaction from the URL string or title alone — the post must stand on real source content.
## Step 3: Classify Content Type
Determine the content type to select the right template:
@ -126,6 +128,7 @@ Verify against quality rules:
- [ ] Character count: 1,200-1,800
- [ ] Topic aligns with expertise areas
- [ ] CTA invites discussion, not just "What do you think?"
- [ ] No corporate buzzwords (leverage, synergy, paradigm shift, thought leader, disruptive, value proposition, ecosystem, holistic approach)
### De-AI / Differentiation Gate
@ -188,6 +191,8 @@ Use WebFetch on each URL. For each, extract:
- **Stance/argument** — what position does the author take?
- **Data points** — any statistics or evidence cited
**If a source can't be extracted** (paywall / JS-heavy / no transcript): apply the Step 2 fallback — drop it if 2+ usable sources remain, otherwise ask the user to paste it or choose another. Never fabricate from a title or URL alone.
### Step 3b: Synthesis Analysis
Analyze across all sources: