diff-profile
Skill509 starsChanged 2 months agoMITView on GitHub
Package your entire Dex system into a shareable DexDiff profile so others can replicate how you work. Use when the user says 'share my whole setup', 'publish my profile'. Not for a single workflow; use `diff-generate`. Not for adopting a whole profile; use `diff-adopt-profile`.
In this repository153 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- .claude/plugins/compound-engineering/CLAUDE.mdCLAUDE.md
- .cursor/rules/search-routing.mdcCursor rule
- anthropic-algorithmic-artSkill
- anthropic-brand-guidelinesSkill
- anthropic-doc-coauthoringSkill
- anthropic-docxSkill
- anthropic-frontend-designSkill
- anthropic-internal-commsSkill
- anthropic-mcp-builderSkill
- anthropic-pptxSkill
- anthropic-slack-gif-creatorSkill
- anthropic-theme-factorySkill
- anthropic-webapp-testingSkill
- anthropic-web-artifacts-builderSkill
- anthropic-xlsxSkill
- atlassian-setupSkill
- backup-nowSkill
- backup-restoreSkill
- backup-setupSkill
- calendar-setupSkill
- commitmentsSkill
- create-mcpSkill
- daily-planSkill
- daily-reviewSkill
- decision-logSkill
- delegate-checkSkill
- dex-backlogSkill
- dex-doctorSkill
- dex-improveSkill
- dex-obsidian-setupSkill
- dex-rollbackSkill
- dex-updateSkill
- dex-whats-newSkill
- enable-semantic-searchSkill
- getting-startedSkill
- goal-backlogSkill
- google-workspace-setupSkill
- granola-setupSkill
- identity-snapshotSkill
- industry-truthsSkill
- initiative-kickoffSkill
- integrate-mcpSkill
- journalSkill
- manage-capabilitiesSkill
- meeting-closeoutSkill
- meeting-prepSkill
- ms-teams-setupSkill
- pipedrive-setupSkill
- pipeline-syncSkill
- process-meetingsSkill
- product-briefSkill
- project-healthSkill
- prompt-improverSkill
- relationship-radarSkill
- reviewSkill
- save-insightSkill
- scrapeSkill
- skill-scoreSkill
- The first 60 shown.
The file
Show its rules and text
We split this skill into the rules it gives an agent
15 rules. About 0 are common, 15 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈15
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 48
prefer fewer, stronger workflows over many thin ones
- Distinctive: line 49
never force a fixed number of workflows
- Distinctive: line 68
Before showing the groupings, explain the judgement clearly:
- Distinctive: line 70
call out any important merges, splits, or dropped candidates when useful
- Distinctive: line 80
Every published workflow should be:
- Distinctive: line 88
If a workflow does not clear this bar, merge it or drop it.
- Distinctive: line 92
When reviewing descriptions for implementation leakage:
- Distinctive: line 94
keep public or commercially available tool names when they help the reader understand the workflow (examples: QMD, Granola, Slack, …
- Distinctive: line 99
if a public tool name stays but may confuse a reader, define or clarify it rather than removing it
- Distinctive: line 103
After generating and saving the profile draft:
- Distinctive: line 106
do not imply that choosing the online path immediately makes it public
- Distinctive: line 107
do not default to editing workflows or the profile overview in the terminal
- Distinctive: line 110
When the user wants to continue, offer this exact shape of next steps:
- Distinctive: line 134
Do not offer "edit one workflow" or "edit overview" as top-level terminal options in the normal flow.
- Distinctive: line 135
If the user explicitly asks to edit locally in the terminal, do that as an exception, not the default.
≈ “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
<!-- Generated from `.claude/skills/diff-profile/SKILL.md` by `scripts/generate-agents-skills.py`. Do not edit. -->What This Command Does
In plain English: Scans your entire vault, identifies your strongest custom workflows, groups them at the right level of granularity, and generates methodology documents for each one - plus an overview of how you use Dex. The result is a complete profile that lives at heydex.ai/diff/your-handle/.
How to run it:
/diff-profileStatus
This command is coming in a future update.
For now, generate individual workflow diffs with:
/diff-generate "my meeting prep workflow"
/diff-generate "my deal review process"
/diff-generate "my weekly planning ritual"Each creates a methodology document that others can adopt individually via /diff-adopt.
What It Will Do (Preview)
- Scan your entire vault — all custom skills, hooks, extensions
- Build candidate workflow clusters based on shared folders, skill references, hook chains, and observed usage sequences
- Score each candidate for:
- distinctiveness — does it represent a meaningfully different job to be done?
- evidence density — is there enough real usage/customisation here to make it credible?
- transferability — would another Dex user understand why this matters and want to adopt it?
- novelty — does it show something differentiated rather than generic Dex usage?
- standalone strength — would this still feel useful if published as its own workflow?
- Merge, split, or drop candidates using these rules:
- merge clusters that feel repetitive, thin, or only make sense together
- split clusters that contain two or more clearly different workflows that would each be valuable alone
- drop clusters that are too generic, too weakly evidenced, or not compelling enough to share
- prefer fewer, stronger workflows over many thin ones
- never force a fixed number of workflows
- Aim for the smallest set of workflows that fully captures the user's differentiated value:
- minimum target: 2-3 workflows if the evidence is thin
- upper bound: 12-15 only if they are genuinely distinct and all clear the quality bar
- otherwise let the model decide the number based on usefulness, not neatness
- Propose boundaries with names and descriptions:
I found [N] strong workflows in your vault: 1. Meeting Prep (3 skills, 1 hook) Job: "Never walk into a meeting cold" 2. Deal Intelligence (4 skills) Job: "Know which deals need attention and why" 3. Weekly Rhythm (3 skills) Job: "Start every week with clarity, end it with evidence" ... Adjust these groupings? [Looks good] [Merge some] [Split one] - Before showing the groupings, explain the judgement clearly:
- say that Dex chose the number of workflows based on strength, not quota
- call out any important merges, splits, or dropped candidates when useful
- make it legible why this set is the right shape for publishing
- Generate a methodology doc for each retained cluster
- Generate an overview narrative: "How I use Dex as a [role]"
- Save everything to the DexDiff profile draft area in the user's vault.
- canonical contract key:
DEXDIFF_PROFILE_DRAFTS_DIR - current default relative path:
04-Projects/DexDiff/beta/profile/
Quality Bar
Every published workflow should be:
- useful to an outsider who has never seen the user's vault
- substantial enough to stand on its own
- clearly different from the other workflows in the profile
- backed by real evidence in the user's setup or usage patterns
- interesting enough to make the profile feel sharp, not padded
If a workflow does not clear this bar, merge it or drop it.
Reviewer Judgement
When reviewing descriptions for implementation leakage:
- keep public or commercially available tool names when they help the reader understand the workflow (examples: QMD, Granola, Slack, Salesforce, Notion, Linear)
- abstract only genuinely internal or private implementation details (examples: company codenames, private scripts, internal project names, vault-specific paths, proprietary dataset names, team-specific naming conventions)
- if a public tool name stays but may confuse a reader, define or clarify it rather than removing it
Handoff After Generation
After generating and saving the profile draft:
- describe it as a local DexDiff draft, not as something already published
- do not imply that choosing the online path immediately makes it public
- do not default to editing workflows or the profile overview in the terminal
- explain that final editing and publish happen on Heydex
When the user wants to continue, offer this exact shape of next steps:
What would you like to do next?
1. Review draft on Heydex
Open the hosted draft/review flow. If a local draft already exists in
`DEXDIFF_PROFILE_DRAFTS_DIR`, reuse that saved draft and the existing machine auth token
to create a fresh review session instead of regenerating. The user can sign in, claim
their profile if needed, edit the workflows and overview on the website, and then choose
when to publish live.
2. Keep local draft for later
Save locally only. Nothing is published. Remind the user that the draft is stored in the
DexDiff profile draft area (`DEXDIFF_PROFILE_DRAFTS_DIR`) and can be regenerated or resumed later.
3. Regroup and regenerate
Rebuild the workflow grouping from scratch using the same underlying evidence.
Explain that this replaces the current draft only after confirmation.
4. Discard this draft
Stop here without opening Heydex or publishing anything.Do not offer "edit one workflow" or "edit overview" as top-level terminal options in the normal flow. If the user explicitly asks to edit locally in the terminal, do that as an exception, not the default.
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.

