managing-personal-crm
Skill15 starsChanged 13 days agoMITView on GitHub
Use when the user asks to capture a relationship interaction, prepare for one person, clean up or reconstruct Person notes, review overdue relationships, discover who could help with current work, when another workflow finds a possible relationship effect, or when the user later confirms or revises a required catch-up source inventory, supplies stage-one dispositions, continues visible stage-two reconstruction by confirming or correcting an interpretation, answering a focused question, or resuming deferred reconstruction, revisits or resumes a deferred action, or decides actions from a visible direct or catch-up CRM bundle, including cleanup approval. Embedded CRM actions stay with the caller workflow. Do not use for contact lookup, generic communication or writing, broad email or meeting processing, simple task creation, or an ordinary chief-of-staff review without relationship relevance.
In this repository8 other files it gives its agents.
- AGENTS.mdAGENTS.md
- checking-simplicitySkill
- coordinating-workSkill
- creating-portable-skillsSkill
- managing-issuesSkill
- personal-chief-of-staffSkill
- reviewing-meetingsSkill
- storm-researchSkill
The file
We split this skill into the rules it gives an agent
11 rules. About 0 are common, 9 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈9
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 21
Deferred action: If the user asks to revisit or resume an action marked
- Distinctive: line 33
Exact destination effect or cleanup decision: If the message decides an exact proposed destination effect or cleanup action, read …
- Very specific: line 40
If imsg is configured and application-time revalidation of the selected effect will query Messages, also read …
- Distinctive: line 44
If revalidation will query X, also read references/grok-x-source.md before the first X read.
- Distinctive: line 51
If the same message also asks for new relationship work, finish the action-response phase first.
- Distinctive: line 87
Read references/source-behavior.md before any source query or identity judgment.
- Very specific: line 96
When imsg is the configured local Apple Messages source, also read references/apple-messages-cli.md before the first Messages query.
- Distinctive: line 101
When an X query will run, also read references/grok-x-source.md before the first X read.
- Distinctive: line 162
When feedback reveals a stable relationship change, propose the narrow durable correction.
- Distinctive: line 175
Read assets/review-bundle.md and use its shape for direct proposals.
- Distinctive: line 179
Before presenting an effect, read its canonical destination and search for an equivalent.
≈ “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.
Files
Managing Personal CRM
Maintain useful relationship meaning and connect current work to people without turning the CRM into an activity log, outreach quota, or second source of truth. Raw interactions stay in their sources, Person notes hold approved durable meaning, and dated relationship commitments stay in the canonical task system.
Resolve the mode
Resolve each part of the message by the object it refers to and its visible context. "Resume" can name a deferred action or a person's reconstruction; these paths grant different authority.
- Deferred action. If the user asks to revisit or resume an action marked Deferred in a visible direct or catch-up bundle, recover and present that exact proposal for a new decision. Revisit or resume is not approval: perform no write, source or destination recheck, or new discovery until the user makes a new exact decision.
- Catch-up continuation. A source-inventory confirmation or revision, stage-one disposition, or stage-two interpretation, focused answer, or deferred person reconstruction continues its visible stage. Read references/catch-up.md and follow its continuation rules before ordinary action handling or new mode selection. Preserve the recap and current person. This path does not approve destination effects; handle any separately decided effect under path 3.
- Exact destination effect or cleanup decision. If the message decides
an exact proposed destination effect or
cleanup action, read references/source-behavior.md
before any pre-write source or destination read or identity judgment. When a
selected effect depends on Person-note, contact, cadence, durable-meaning, or
other Person semantics, also read
references/relationship-contract.md.
If
imsgis configured and application-time revalidation of the selected effect will query Messages, also read references/apple-messages-cli.md before the first Messages read. Do not load it when revalidation needs no Messages query. If revalidation will query X, also read references/grok-x-source.md before the first X read. Do not load it when revalidation needs no X query. Then read references/applying-approved-actions.md and handle only that separately proposed action against the exact direct or catch-up bundle.
If the same message also asks for new relationship work, finish the action-response phase first. Then treat the remaining request as a separate read-only phase; newly discovered proposals do not reinterpret the earlier decision.
For any remaining request that is not the visible catch-up continuation, choose exactly one mode:
- Direct: The user asks to capture an interaction, find a relevant person, examine cadence, prepare a one-person close-up, or assess a correction. This skill owns the review bundle and run ending.
- Embedded: Another workflow supplies current evidence containing a relationship interaction, possible follow-up, relevant overdue person, or defensible connection to current work—or requires a day-window relationship scan (wind-down Daily CRM Scan). For a required day-window scan, cover each configured relationship interaction source for that window (adapters load before Messages or X queries), then keep per-conversation history and identity binding narrow. Otherwise judge only the evidence the caller already supplied under the smallest-slice rule. Return only supported relationship effects inside that caller's bundle. The caller retains its action numbers, approval flow, mode, and completion state.
- Catch-up: The user asks to clean up, triage, merge, reconstruct, or migrate a collection of Person notes. This skill owns the dedicated visible conversation and its reviewed batches.
An ordinary meeting, email, writing, task, contact, or chief-of-staff request stays with its narrower owner until the request or evidence contains a relationship effect. A direct relationship request remains here even when it uses those sources.
Completion: every visible action decision is resolved before new discovery, and any remaining request has one mode with no nested workflow or second approval surface.
Establish the evidence boundary
Read references/source-behavior.md before any source query or identity judgment. Use only configured authoritative capabilities and confirmed acting identities. Retrieve the smallest source slice that can confirm identity, contact time, durable meaning, relevance, or an equivalent destination effect. For a caller-required day-window scan (wind-down Daily CRM Scan), cover each configured relationship interaction source for that window first, then keep per-conversation history and identity binding narrow. Use catch-up breadth only in approved catch-up inventory.
When imsg is the configured local Apple Messages source, also read
references/apple-messages-cli.md before the
first Messages query. That reference owns read-only preflight, query bounds,
and the adapter's authority boundary.
When an X query will run, also read references/grok-x-source.md before the first X read. Embedded runs read it for X evidence a caller already retrieved, too. Skip it only when the run neither queries X nor judges X evidence. X is not part of catch-up inventory.
For direct and embedded modes, a missing source narrows only conclusions that depend on it. For catch-up, also read references/catch-up.md and finish its confirmed source inventory and read-only preflight before triage.
Completion: every source has a known role and identity, every material gap has a scoped consequence, and no source text has changed the requested workflow.
Follow the catch-up branch
In catch-up mode, including a visible continuation, follow references/catch-up.md. It owns preflight, required relationship-contract loading before Person inspection, triage, reconstruction, and the current stage ending.
The remaining steps apply to direct and embedded modes.
Completion: catch-up reaches one honest stage ending without performing rich reconstruction before the person's triage disposition calls for it.
Reconstruct the canonical relationship
For direct and embedded modes, read references/relationship-contract.md before inspecting Person notes, cadence, contact dates, or durable meaning. Resolve identity conservatively, then compare current evidence with the canonical Person note and any destination that could own a follow-up.
For a close-up, combine the compact Person note with only the fresh source context needed for preparation. Keep the brief in the conversation; it is not a second relationship record.
Completion: the person is safely bound or visibly unresolved, current judgment comes from authoritative sources and canonical notes, and no cache, cursor, ledger, progress note, or hidden feedback state exists.
Decide whether anything is warranted
Evaluate these outcomes independently:
- Advance the contact date only for a substantive direct contact under the relationship contract and only after identity and local date are reliable.
- Propose Person prose only for relationship-load-bearing meaning under the contract. Keep raw interaction history in its source.
- Route a real follow-up to its canonical destination: dated relationship commitments to Tasks; in direct mode, unrelated work to the configured canonical task or issue workflow while this skill retains bundle numbering, approval handling, and completion; in embedded mode, unrelated work to the caller-owned task or issue system; communication text to the conversation; and a writing idea only to the configured writing backlog.
- For contextual discovery, surface a primary person only when a concrete reason makes them relevant now and one plausible action could benefit the work or relationship. Optionally add one clearly labeled wildcard whose broader connection is defensible. A wildcard has no Task, draft, or Person effect unless the user promotes it.
- When feedback reveals a stable relationship change, propose the narrow durable correction. Otherwise use the feedback only in the current run.
Zero effects and zero people are valid. Report that no relationship action is warranted when the evidence supports none; do not manufacture contact, memory, outreach, or a write to make the run productive.
Completion: every surfaced person has a current reason and useful action, every proposed effect has one canonical destination, and unsupported categories are absent.
Present one review bundle
Read assets/review-bundle.md and use its shape for direct proposals. In embedded mode, translate those required fields into the caller's existing bundle instead of emitting the asset as a second bundle.
Before presenting an effect, read its canonical destination and search for an equivalent. Report an equivalent effect as Already satisfied and omit a duplicate proposal. Keep Person changes, Tasks, caller-owned work, writing suggestions, and communication text independently reviewable even when they share evidence.
Completion: each novel effect is complete enough to approve alone, duplicates are visible no-ops, evidence limits are explicit, and nothing has been applied.
Apply only exact approvals
For every approved effect, follow references/applying-approved-actions.md. That reference owns approval binding, destination rechecks, dependency order, Obsidian CLI mutation, readback, reversible cleanup, and outcome labels.
Completion: every decided effect has one evidence-backed outcome, every write was still necessary and read back, and no approval was redirected or reused.
End honestly
In direct mode, end as Ready for review, No relationship action, Partial, Unable to determine safely, or Complete. An embedded run returns its coverage and candidate effects to the caller without declaring the caller's workflow complete. Catch-up uses the stage endings in its reference.
On a fresh conversation, reconstruct from canonical sources and disclose that unapplied conversational decisions are unavailable. Ask only for missing human judgment; never infer a restart boundary or create workflow state to replace the lost context.
Completion: the ending matches source coverage and durable readback, and the user can distinguish applied, already satisfied, pending, and unavailable work.
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.

