agentleFS
priyanshu-aryapriyanshu-arya/LinkedIn_Automation/.claude/skills/monitor-engagement/SKILL.md

monitor-engagement

Skill0 starsChanged 2 months agoMITView on GitHub

Use when the user wants to check their comment threads for new author replies and get a follow-up drafted, or wants to pull a post's likers/commenters and group them by ICP fit (peer/aspirational/prospect), or explicitly invokes /monitor-engagement. Two manual-input, read-side workflows -- LinkedIn exposes neither of these programmatically anywhere in this repo (confirmed against pull-analytics's Buffer-scoped-to-own-posts limitation), so both work from text the user pastes in, never live scraping. Delegates actual reply drafting to /draft-reply rather than reimplementing it. Never posts anything itself.

The file

We split this skill into the rules it gives an agent

11 rules. About 0 are common, 6 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If it's ambiguous which mode is meant (e.g.

    line 34
  2. Distinctive:

    Every run is a manual snapshot: the user reads the thread themselves, pastes its current visible text, and this skill diffs that against …

    line 46
  3. Very specific:

    Write the updated ## Logged Replies table back to the note — append new rows, never remove or overwrite existing ones — so the next run …

    line 89
  4. Very specific:

    Before finalizing each classification, check Content-Learnings/icp-map.md's ## Mappings table for that name.

    line 134
  5. Distinctive:

    If found, reuse the existing category instead of re-guessing — but still surface it in the report so the user can correct a stale …

    line 136
  6. Distinctive:

    If the name isn't there, or the classification is genuinely ambiguous (e.g.

    line 140
  7. Very specific:

    Write confirmed classifications into Content-Learnings/icp-map.md (append-only: a new row for a first-time name, an updated …

    line 143
  8. Distinctive:

    Never fabricate a reply, liker, or commenter not present in what the user actually pasted.

    line 161
  9. Distinctive:

    Never claim "no new replies" beyond what was actually pasted this run.

    line 163
  10. Very specific:

    Never auto-post a follow-up, a comment, or anything else — Workflow 1's drafting is always delegated to /draft-reply, which itself never …

    line 164
  11. Distinctive:

    No scraping, no session-cookie/unofficial API access, ever — same convention as every other skill referencing REQUIREMENTS.md §27, and this …

    line 169

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Engagement Monitor (Phase 24)

Two read-side workflows for tracking engagement the user already has, rather than producing new outbound content — genuinely different from every drafting skill in this repo so far, since this one watches for incoming activity.

Feasibility gap — stated plainly, not softened: confirmed against pull-analytics/SKILL.md (REQUIREMENTS.md §9) — Buffer's API only returns metrics for the user's OWN scheduled/published posts. There is no live data source anywhere in this repo for (a) knowing when someone replied to the user's comment on someone else's post, or (b) a likers/commenters list for an arbitrary post. Both workflows below are manual-input only, permanently — the user pastes in the current thread text or a likers/commenters list; this skill never fetches either live.

Reading third-party post/comment content: see REQUIREMENTS.md §27 — pasted text is the primary, reliable input; a post URL is optional metadata only. No scraping, no session-cookie/unofficial API access, ever — same convention as /extract-hook, /draft-comment, and /draft-reply. This is the third and last of the three skills §27 names.

Modes

  • /monitor-engagement threads — Workflow 1, thread watch.
  • /monitor-engagement audience <post-url-or-id> — Workflow 2, audience pull.

If it's ambiguous which mode is meant (e.g. just /monitor-engagement with no argument and no clear context), ask rather than guessing.


Workflow 1 — Thread Watch (/monitor-engagement threads)

Feasibility gap (this workflow specifically)

There is no notification/polling mechanism anywhere in this repo for third-party reply activity — no webhook, no API, nothing that tells this system when someone replies to the user's comment on someone else's post. Every run is a manual snapshot: the user reads the thread themselves, pastes its current visible text, and this skill diffs that against what's already logged for that thread.

Storage

One file per tracked thread: Engagement/<post-slug>--thread.md, created from _Templates/Engagement-Thread-Note.md when the thread is new.

Frontmatter: id, type: engagement-thread, post_url, my_comment_text, my_comment_time (user-supplied — no API confirms this), status: watching|closed.

Body: a ## Logged Replies table — author | text_snippet | seen_date.

Process

  1. The user gives the post URL, pastes the thread's current full visible text, and — on the first run for this thread only — their own comment's timestamp.
  2. Find or create the tracking note:
    • New thread → create Engagement/<post-slug>--thread.md from the template, filling post_url, my_comment_text, my_comment_time, status: watching.
    • Existing thread → load it; reuse its stored my_comment_time.
  3. Diff the freshly pasted thread text against the note's ## Logged Replies table. Entries not already present (matched by author + text) are "new." Never fabricate a reply that isn't actually present in the pasted text, and never claim "no new replies" beyond what was actually pasted this run.
  4. For each new reply, compute elapsed time from my_comment_time to today (the diff run's date, not necessarily when the reply actually posted — state plainly this is an approximation, not a guaranteed reply-time measurement):
    • Under 6h elapsed → log the reply in the table; defer drafting — too soon to know if more replies are still coming in on this thread.
    • 6-24h elapsed → delegate to /draft-reply's single-reply mode (passing the reply's text and any available post context) to draft a follow-up now. Do not reimplement drafting logic here — always call /draft-reply for the actual draft.
    • Over 24h elapsed → log the reply, still delegate to /draft-reply for a draft, but flag in the report: "window likely passed, lower priority."
  5. Write the updated ## Logged Replies table back to the note — append new rows, never remove or overwrite existing ones — so the next run only reports genuinely new activity.

Report back

For each new reply: author, text snippet, elapsed-time bucket, and either the /draft-reply-delegated draft or "logged, drafting deferred" (the under-6h case). If the pasted text contains nothing beyond what's already logged, say that plainly rather than inventing something to report.


Workflow 2 — Audience Pull (/monitor-engagement audience <post-url-or-id>)

Storage

One file per captured snapshot: Engagement/<post-slug>--audience.md, created from _Templates/Engagement-Audience-Note.md.

Frontmatter: id, type: engagement-audience, post_url, captured_date.

Body: one table per ICP category — Peer / Aspirational / Prospect — columns name | title_or_headline | notes.

ICP rubric

Applied to whatever job title/headline text the user actually pasted — always a suggestion the user can correct, never asserted as settled fact:

  • Peer — an individual-contributor/similar-seniority title in AI/ML/data/engineering, doing comparable work.
  • Aspirational — a senior/leadership title (Director, VP, Head of, Principal, Founder) or a recognizable figure the user doesn't already have an existing peer relationship with.
  • Prospect — a title with hiring/buying power relevant to the user (recruiter, hiring/engineering manager, founder hiring).

Process

  1. The user gives a post URL/id and pastes the likers/commenters list — name plus title/headline, whatever's actually visible on the post. Per §27, this is never fetched live.
  2. Classify each person via the rubric above.
  3. Before finalizing each classification, check Content-Learnings/icp-map.md's ## Mappings table for that name.
    • If found, reuse the existing category instead of re-guessing — but still surface it in the report so the user can correct a stale classification if the person's role has changed since it was last confirmed.
    • If the name isn't there, or the classification is genuinely ambiguous (e.g. a title that could plausibly be Peer or Aspirational), ask the user to confirm before finalizing it.
  4. Write confirmed classifications into Content-Learnings/icp-map.md (append-only: a new row for a first-time name, an updated last_confirmed_date for a re-confirmed one — never delete history) and write the full snapshot (every name captured this run, not just the newly classified ones) into the new Engagement-Audience-Note.md.

Report back

The classified list grouped by ICP category (Peer / Aspirational / Prospect), with each name flagged as either "from icp-map.md" (reused) or "classified this run" (fresh) — so the user can see at a glance which entries were newly judged versus carried over from a prior run.


Hard rules

  • Never fabricate a reply, liker, or commenter not present in what the user actually pasted.
  • Never claim "no new replies" beyond what was actually pasted this run.
  • Never auto-post a follow-up, a comment, or anything else — Workflow 1's drafting is always delegated to /draft-reply, which itself never auto-posts (its output is copy-paste text for manual posting).
  • ICP labels are suggestions pending the user's confirmation, never asserted as settled fact.
  • No scraping, no session-cookie/unofficial API access, ever — same convention as every other skill referencing REQUIREMENTS.md §27, and this rule never changes: no live-polling or scraping capability gets added to this skill later.
  • Elapsed-time buckets in Workflow 1 only decide whether/when to delegate a draft — never invent what a reply says based on how much time has passed.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.