agentleFS
Ethan9123Ethan9123/wx-life-analysis/.claude/skills/subtext-reading/SKILL.md

subtext-reading

Skill0 starsChanged 6 months agoApache-2.0View on GitHub

Read the right-now state of a specific WeChat conversation — is it warm, cooling, or disengaging — and decide the correct next move given who currently holds the ball. Activate when the user asks "is she/he cooling on me?", "what's the subtext of that reply?", "should I ping them again?", "am I being too pushy?", or after a refresh when reviewing what changed. Walks the last 20-50 messages in people/<slug>/chat.md and outputs a 5-line block (state / ball-in-court / last move / key signal / do now) into the "当前状态" section of profile.md. Never hedges — picks exactly one of six states (🔥 hot · 🟢 warm · 🟡 mild cool · 🟠 cooling · 🔴 disengaging · ⚫ gone).

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

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

    When this skill is loaded, you are diagnosing where this specific conversation is at this specific moment — orthogonal to who the person is …

    line 8
  2. Very specific:

    Full methodology: read docs/subtext-reading.md.

    line 10
  3. Distinctive:

    Always 5 lines, no prose paragraph:

    line 44
  4. Very specific:

    Update the "当前状态" section in people/<slug>/profile.md body (not frontmatter — state changes too often).

    line 55

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Subtext Reading Skill

When this skill is loaded, you are diagnosing where this specific conversation is at this specific moment — orthogonal to who the person is in general (that's the mbti-analysis skill).

Full methodology: read docs/subtext-reading.md. It defines the 9-signal checklist, the 6 states, the ball-in-court × state action matrix, and the common false positives.

Quick reference

The 6 states (pick ONE)

StateMeaningDefault action
🔥 hotactive, fast back-and-forth, energy matchedcontinue at current cadence
🟢 warmsteady, normal pace, no negative signalsmaintain
🟡 mild coollatency stretched, length shrinking, no hard signalhold. do not ping.
🟠 cooling2+ signals from different categoriespull density down, don't try to "fix"
🔴 disengagingexplicit redirect / sticker-only / >48h silence on a clear ballstop. acknowledge if anything, no new topics
⚫ gone>7 days silent, no ack on last 2 messagesdon't chase. one neutral close max.

The 9 signals (find 2+ before calling cooling)

Latency: (1) reply gap spike, (2) conversational stop · Length: (3) length collapse, (4) stickerization · Content: (5) hard redirect, (6) soft non-answer, (7) topic never gets hook, (8) explicit boundary · Meta: (9) 🚨 critique of your style not the topic — "you sound like a bot" — this one's critical.

Ball-in-court × state action matrix

StateBall in their courtBall in my court
🟡 mild coolwait, do not double-pingreply normally, no escalation
🟠 coolingwait minimum 3× their normal gapreply low-stakes, no big swing
🔴 disengagingdo not ping unpromptedreply minimal, no question, no hook

The most common Ethan-style mistake (per the methodology doc): ball in their court + mild cool → user sends a second "check in" message → converts mild cool into cooling. Call this out explicitly when relevant.

Output format

Always 5 lines, no prose paragraph:

[name]
state: <one of 6 emoji + label>
ball: me / them
last move: <date> · <what happened in one line>
key signal: <which signals matched, e.g. "length collapse + soft non-answer">
do now: <one concrete action, e.g. "hold. no ping. wait 3+ days.">

Update the "当前状态" section in people/<slug>/profile.md body (not frontmatter — state changes too often).

Hard rules

  1. Pick one state. No hedging. "warm-to-mild-cool" defeats the purpose.
  2. 2+ signals required to call cooling. One signal = noise (e.g., 6h reply gap = could be sleeping).
  3. Don't paste chat.md content outside the local session.
  4. For distant relationships (>1 month no contact), this skill is calibrated wrong — use mbti-analysis instead.

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.