sergebulaev/linkedin-skills / .codex-marketplace/linkedin-skills/skills/linkedin-thread-monitor
linkedin-thread-monitor
VerifiedTrack which of your LinkedIn comments earned author replies. Flags the 6-24h warm-reply window where thread momentum peaks, classifies threads as hot/warm/cool/dormant, and routes warm ones to linkedin-reply-handler for follow-up drafts. Powered by Apify, no LinkedIn login. Triggers on "what threads need follow-up", "author replied", "monitor my comments". Not for analyzing likers on a post (use linkedin-engager-analytics).
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 5.8 KB |
| references/output-spec.md | markdown | 1.2 KB |
| references/thread-timing.md | markdown | 3.7 KB |
references/thread-timing.md
View on GitHub# Thread Timing Matrix
## Thread stage classification
| Time since user's comment | Time since last reply | Stage | Priority |
|---|---|---|---|
| <6h | any | Watch (author may still reply) | Low — check back |
| 6-24h | author replied <2h ago | **Hot** — respond within 90 min | HIGH |
| 6-24h | author replied 2-12h ago | **Warm** — respond within 2h | HIGH |
| 6-24h | no author reply | Cold — skip | — |
| 24-72h | author replied recently | Cool — respond within 4h | Medium |
| 24-72h | no author reply | Dormant | — |
| >72h | any | Dormant — switch to DM | Medium (if inbound-quality) |
## The warm-reply window explained
Real example from 2026-04:
- 14:27 UTC: Serge posted comment on a CEO's post ("moat moved from tools to taste")
- 12:06 UTC next day (~22h later): the author replied personally ("How are you building that conviction muscle with your team?")
- 16:24 UTC that day (~28h after original comment, ~4h after the author's reply): Serge replied with his answer
This is the exact window the skill targets. Miss it by 12+ hours and the reply lands in a dormant thread where the author doesn't get the notification prominently.
## First 60 min on own posts
Different metric — how fast the USER replies to comments on their own posts:
- Target: every comment replied to within 5-15 min during first 60 min
- Each reply within 90 min fires ~90% boost on that thread
- 3+ substantive comments in first 30 min = second algo push
## Engagement half-life
- **0-6h:** 70% of all eventual reactions/comments happen here
- **6-24h:** 25% — the long tail
- **24-72h:** 5% — trickle
- **>72h:** essentially dead (<1% of eventual engagement)
## Rule: when thread dies, switch to DM
If a thread is dormant (>72h since last turn) but the counterpart was high-quality, don't reply in thread — the post won't surface their notification. Instead, draft a DM:
```
[Name] — circling back on our thread about [specific topic from thread].
[Your one new thought or data point].
Worth a 15-min conversation? Tuesday or Thursday this week if yes.
```
The DM should reference the thread specifically, not be a generic pitch.
## Anti-patterns
- Chaining 3+ replies under one top comment (looks like thread hijack)
- Replying after 72h in the thread itself (low visibility, looks desperate)
- Generic "catching up on this thread" without a new thought
- DMing before the public thread closes naturally (skips the earned step)
- Replying to replies OF replies (LinkedIn flattens — it doesn't nest that deep)
## Publishing-adjacent timing windows (own posts)
| Phase | Window | Action |
|---|---|---|
| Warm-up | 15 min **BEFORE** publishing | Leave 3-5 substantive comments on others' posts |
| Critical | First 30 min AFTER publishing | Reply to every comment within minutes; distribution contracts if dead |
| Seeding | 15-30 min after posting | Leave 3-5 bonus comments on your own post to create thread depth |
| Visibility bump | Reply within 1st hour | +35% visibility lift (author-reply signal) |
## Peer engagement (safe pattern, not a pod)
A **safe peer group** is 5-8 people in adjacent fields who actually read each other's work and comment only when they have something substantive to say.
Distinguishes from pods by:
- Varied timing (no fixed daily slot)
- Varied commenters per post (not the same 6 people every time)
- Comment substance >10 words, with new angles
- No reciprocity obligation
Pod detection catches:
- Same accounts engaging at the same clock minute daily (e.g., 9:01 AM)
- 15+ comments landing within a 90-second window
- Identical like/comment pattern across every post
Real penalty observed: one creator dropped from 8,500 to 340 impressions overnight after pod detection. Recovery: 6-8 weeks.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| deprecation-and-migrationaddyosmani/agent-skillsManages deprecation and migration. Use when removing old systems, APIs, or features. Use when migrating users from one implementation to another. Use when migrating a database schema in production, such as renaming or dropping a column without downtime (expand/contract). Use when deciding whether to maintain or sunset existing code. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| performance-optimizationaddyosmani/agent-skillsOptimizes application performance across frontend, backend, queries, and databases. Use when performance requirements exist, when you suspect performance regressions, when Core Web Vitals or load times need improvement, when N+1 query patterns need fixing, or when profiling reveals bottlenecks. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| agent-cost-reportthedotmack/claude-memBelievable agent cost report for any period, default the last 7 full days PT, not counting today. Measured tokens from Claude Code transcripts priced at OpenRouter list prices (ESTIMATED), measured provider spend when a sanctioned source exists, note-taker cost separate, Timing-style HTML/PDF plus report.json, line-items.csv, evidence.json. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| agent-cost-reportthedotmack/claude-memBelievable agent cost report for any period, default the last 7 full days PT, not counting today. Measured tokens from Claude Code transcripts priced at OpenRouter list prices (ESTIMATED), measured provider spend when a sanctioned source exists, note-taker cost separate, Timing-style HTML/PDF plus report.json, line-items.csv, evidence.json. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |