record-keeper
Skill33 starsChanged 11 days agoMITView on GitHub
Record Keeper. Logs every email and call for you, so you never update a CRM by hand again. Reads your inbox through the Gmail connector, matches every message to a contact on the board, and logs it (opening, follow-up or reply) into closer/pipeline.json; logs calls with their notes saved to closer/calls/; shows the full timeline with anyone; and lists new addresses writing in that are not contacts yet. Use when the user says "log this", "log my emails", "sync my inbox", "update the CRM", "what's my history with X", "save the call notes", or at the end of every call and every morning.
In this repository10 other files it gives its agents.
- contract-writerSkill
- follow-upSkill
- forecasterSkill
- pipelineSkill
- proposal-writerSkill
- prospectorSkill
- researcherSkill
- sales-coachSkill
- schedulerSkill
- win-backSkill
The file
We split this skill into the rules it gives an agent
10 rules. About 0 are common, 6 are distinctive, and 4 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.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 36
Read only those threads. Never read other mail for this.
- Very specific: line 37
Write what you found to a sync file: a JSON list of {"date": "YYYY-MM-DDTHH:MM", "from": "...", "to": ["..."], "subject": "...", "text": …
- Very specific: line 42
If a reply says no or asks to stop, tell the user and run /follow-up's reply for that person so every skill stops.
- Very specific: line 46
No Gmail connector? The user can paste or forward the emails, and you log them with log.py log.
- Very specific: line 50
After every call, write the notes in plain lines: what they need, what they said about price, what was agreed, the next step and its date.
- Distinctive: line 65
Never logs an outgoing message the user did not send or approve.
- Distinctive: line 66
Never reads mail beyond the threads with known contacts, and never stores attachments.
- Distinctive: line 67
Never edits or deletes what is already logged: a correction is a new note.
- Distinctive: line 75
Stop at the first "no" or any sign of disinterest.
- Distinctive: line 76
Never invent what was said.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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.
Files
record-keeper
The Record Keeper keeps the board true without anyone typing into it. Every email in and out and
every call ends up in closer/pipeline.json (the schema /pipeline owns), once.
${CLAUDE_SKILL_DIR} below is this skill's folder (where this SKILL.md lives).
python3 "${CLAUDE_SKILL_DIR}/log.py" sync --emails closer/inbox-2026-10-27.json
python3 "${CLAUDE_SKILL_DIR}/log.py" unknown --emails closer/inbox-2026-10-27.json
python3 "${CLAUDE_SKILL_DIR}/log.py" log --contact dana-harbor-lane --channel email --direction in \
--subject "Re: Quick idea for Harbor Lane Dental" --text "Sorry, crazy week! Thursday at 2 works." \
--at 2026-10-27T11:48
python3 "${CLAUDE_SKILL_DIR}/log.py" call --contact dana-harbor-lane --minutes 24 \
--notes-file /tmp/harbor-lane-notes.md --at 2026-10-29T14:00
python3 "${CLAUDE_SKILL_DIR}/log.py" timeline --contact dana-harbor-laneSyncing email
- With the Gmail connector, search the user's mail since the last sync for messages to or from
any contact's address (
from:dana@harborlanedental.com OR to:dana@harborlanedental.com ...). Read only those threads. Never read other mail for this. - Write what you found to a sync file: a JSON list of
{"date": "YYYY-MM-DDTHH:MM", "from": "...", "to": ["..."], "subject": "...", "text": "..."}with the text trimmed to the new part of each message (no quoted history). log.py sync --emails <file>. It skips anything already logged, so running it twice is safe. Email the user sent from their own inbox is logged with approval "sent by you from your inbox".- If a reply says no or asks to stop, tell the user and run
/follow-up'sreplyfor that person so every skill stops. Do not just log it. log.py unknownlists addresses that wrote in but are not contacts. Ask the user before adding any.
No Gmail connector? The user can paste or forward the emails, and you log them with log.py log.
Logging a call
After every call, write the notes in plain lines: what they need, what they said about price,
what was agreed, the next step and its date. Then log.py call. The notes file lands at
closer/calls/<date>-<company>.md and a one-line summary goes on the board. If /sales-coach
ran on the call, use its review output as part of the notes.
Example from the reel
Dana's thread in the reel, logged with no typing: the opening on Mon Oct 19 at 9:02 AM, follow-ups
on Oct 21, 23 and 27, her reply on Tue Oct 27 at 11:48 AM ("Sorry, crazy week! Thursday at 2 works."),
then the call on Thu Oct 29: "Call, 24 min. Loved the site. Price objection on $2,500; agreed to the
$399/mo plan, no setup fee. Wants to start Monday." timeline --contact dana-harbor-lane prints all of
it. (Illustrative: Dana and Harbor Lane Dental are fictional.)
What it never does
- Never logs an outgoing message the user did not send or approve.
- Never reads mail beyond the threads with known contacts, and never stores attachments.
- Never edits or deletes what is already logged: a correction is a new note.
Consent and truth
- Nothing is sent, booked or signed without the user's explicit yes for that exact message, invite or document. This skill only records; it never sends anything.
- Calls: tell everyone on the call that captions or notes are on, before you start. Some places (for example California, Florida, Washington, and many countries) need every person's consent to record.
- Stop at the first "no" or any sign of disinterest. A request to stop is honoured forever, across every skill.
- Never invent what was said. Notes are what happened, in plain words.
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
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.

