Claude, Client Comms Log.
Uses Claude's client-comms-log skill to show that turning a call or email into a structured log entry is a fixed template, not a stored memory of client history.
Claude has a skill called client-comms-log, and it's tempting to assume Claude has started remembering your clients across conversations. It hasn't. A Claude Skill is a folder holding one file, SKILL.md, that Claude reads before it starts and then follows step by step, in order. Delete the folder and Claude forgets no client history, because nothing was ever stored between runs to forget.
What the skill actually is
Client-comms-log is a folder holding a single SKILL.md with instructions, in plain language, for turning a call or an email into a structured log entry. Claude reads that file top to bottom and works through it step by step in the order written, branching only where the file itself says to. Nothing about this involves an ongoing record of who a client is or what was discussed last time; each run is a fresh application of the same structure.
Specification, not memory
This makes a skill a specification, not a memory. The payoff is consistency: the same log structure comes out of every entry, date, contact, channel, summary. The limit is scope: anything outside those written steps, and Claude is off the map. Client-comms-log never gave Claude a memory of your clients. It guarantees only that every time the skill runs, Claude reads the same file and produces the same log structure, which is a much narrower and more accurate claim than "Claude remembers your clients."
Why deleting the skill changes nothing about memory
If the skills folder were deleted, Claude wouldn't forget a single client, because there was nothing stored to forget in the first place. What disappears is one specific logging routine, a way of turning a conversation into a structured entry, not an accumulated relationship history.
What a log entry does and doesn't prove
Watching Claude produce a clean log entry doesn't prove it understood the call. A file's structure can be followed to the letter on a conversation that was never really grasped in any deep sense. Watching Claude produce a messy entry doesn't prove the skill is broken either; it may just be a call whose content doesn't fit the file's structure well. Neither outcome proves comprehension or its absence.
Trying it yourself
Pick one record you create the same way every time. Ask Claude to write a SKILL.md for it, plain language, ordered steps, then have it read the file back to you and walk through exactly what it will do before doing it. Running this against a record you actually produce repeatedly shows the mechanism directly.
Key takeaways
- Claude's client-comms-log skill is one plain-language SKILL.md describing how to turn a call or email into a structured log entry.
- Claude follows the file's steps in a fixed order, branching only where the file explicitly says to.
- A skill is a specification for structure, not a memory system; deleting it loses a routine, not stored client history.
- The same log fields, date, contact, channel, summary, come out on every entry the skill produces.
- A clean or messy log entry proves nothing on its own about whether Claude understood the underlying call.
Who this is for
Client-facing teams who want consistent communication logs without assuming Claude retains memory across sessions, and anyone curious about what a logging-style Claude Skill actually contributes.
Chapters
- 0:00The naive framing: "Claude remembers every client call"
- 0:12Sounds like a relationship upgrade
- 0:22Broken, with a case — delete the folder, nothing is forgotten
- 0:32The anchor: one file — client-comms-log's single SKILL.md
- 0:46Read it, then follow it in order
- 0:55Spec, not memory — the payoff and the limit
- 1:06The anchor returns — same file, every run
- 1:18Both directions — neither one is proof
- 1:33Carry-out
- 1:43Your turn
- 1:55Outro
Full transcript(auto-generated, with timestamps)
The naive framing: "Claude remembers every client call"
[0:00]You guess Claude remembers every client call once it uses the client comms log skill. It doesn't. It was handed a file of instructions. Client comms log just names one such file. Let's see what's actually inside it. Here Claude has a
Sounds like a relationship upgrade
[0:13]Client comms log skill and it sounds like a relationship upgrade, like it now keeps a running memory of every client across every conversation. But nothing
Broken, with a case — delete the folder, nothing is forgotten
[0:22]Is stored between runs. Delete the skills folder and Claude doesn't forget a single client. There was nothing to forget. It just stops following that one logging routine. Here's what a skill
The anchor: one file — client-comms-log's single SKILL.md
[0:32]Actually is, one file. Client comms log for instance is a folder holding a single skill. MD instructions for turning a call or email into a structured log entry written in plain language Claude reads before it starts.
Read it, then follow it in order
[0:46]Claude reads that file top to bottom and works through it step-by-step in the order it's written, no branching unless the file itself says branch. That makes
Spec, not memory — the payoff and the limit
[0:55]A skill a specification, not a memory. The payoff, the same log structure every entry date, contact channel summary. The limit, anything outside those written steps and Claude is off the map. So
The anchor returns — same file, every run
[1:06]Client comms log never gave Claude a memory of your clients. It just guarantees that every time client comms log runs, Claude reads that same file and produces the same log structure. That's the whole trick. Watching Claude
Both directions — neither one is proof
[1:19]Produce a clean log entry doesn't prove it understood the call. A file can be followed to the letter on a conversation it never really grasped. And watching it produce a messy entry doesn't prove the skill is broken. It may just be a call the file structure doesn't fit. A skill
Carry-out
[1:33]Doesn't give Claude a memory of your clients. It's a file of steps Claude reads before it starts. So every log entry comes out with the same structure every time. Your turn. Paste this into Claude. Pick
Your turn
[1:44]One record I create the same way every time. Write me a skill MD for it, plain language ordered steps, then read it back to me and walk me through exactly what you'll do before you do it. Claude
Outro
[1:55]Client comms log, Liam in for bear.
