agentleFS
Pearl1531Pearl1531/Archetype-talk/.claude/skills/dovetail-sync/SKILL.md

dovetail-sync

Skill2 starsChanged 49 days agoMITView on GitHub

Connect a Dovetail workspace over its official MCP as a read-only research source and map interviews/highlights/insights into Transcripts/Signals/Evidence with provenance. Use when the user mentions Dovetail or wants to sync/refresh research from it.

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 6 are distinctive, and 2 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.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Very specific:

    Read-only: Use only Dovetail's search_* / get_* / list_* tools.

    line 17
  2. Very specific:

    Provenance always: Every imported file records where it came from (source: dovetail, dovetail_id, dovetail_url) so it's traceable and …

    line 18
  3. Distinctive:

    Read-only tools only: If a Dovetail write/create tool is available, do not use it.

    line 37
  4. Distinctive:

    No fabrication: Import only what Dovetail actually returns.

    line 39
  5. Distinctive:

    Never writes to Dovetail or changes Dovetail content.

    line 62
  6. Distinctive:

    Never commits API tokens or puts secrets in the repo.

    line 63
  7. Distinctive:

    Never overwrites a hand-written file without showing a diff and confirming.

    line 64
  8. Distinctive:

    Never turns a synthetic persona chat into a Dovetail record (that pipeline is one-way: Dovetail → us).

    line 65

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

dovetail-sync — Connect Dovetail as a research source (MCP)

Safely connect this project to a Dovetail workspace over the official Dovetail MCP server, then pull research into our schema (Transcripts / Signals / Evidence) with full provenance. Dovetail is a read-only source of knowledge — it never dictates or changes our folder structure.

Core principle — direction of truth

Dovetail (source)  ──read-only──▶  our graph (canonical)
  • Our schema is canonical. Dovetail data is mapped INTO Transcripts/, Signals/, Evidence/ — never the other way around.
  • Read-only. Use only Dovetail's search_* / get_* / list_* tools. Never call create/update tools; never write anything back to Dovetail.
  • Provenance always. Every imported file records where it came from (source: dovetail, dovetail_id, dovetail_url) so it's traceable and re-syncable.
  • Grounding rules still apply. Imported Signals keep verbatim quotes + real dates; Evidence gets a retrieved: date; the 3-month freshness rule holds (see ai-persona/references/levels.md).

When to use

  • The user wants to connect Dovetail, or refresh data from it.
  • The user has interviews/highlights/insights in Dovetail and wants them as Signals/Evidence here.

Trigger

/dovetail-sync            ← full flow: ensure connection → discover → map → confirm → write
/dovetail-sync --setup    ← only walk through the safe connection setup
/dovetail-sync <project>  ← sync a specific Dovetail project

Safety rules (read before doing anything)

  1. Never handle secrets in chat. Do NOT ask the user to paste an API token into the conversation, and never type one into a file. Prefer OAuth / first-party login (browser). If a token is unavoidable, the user configures it in their MCP client's own secret store / env — see setup.md. Tokens must never be committed to the repo.
  2. Read-only tools only. If a Dovetail write/create tool is available, do not use it.
  3. Confirm before writing our files. Show the user a plan (what will become which file) and get a yes before creating/updating anything in Transcripts/, Signals/, Evidence/.
  4. No fabrication. Import only what Dovetail actually returns. Never invent quotes, numbers, dates, or participants. If a Dovetail item is thin, import it thin.
  5. Don't restructure our data to match Dovetail. Map into our types; if something doesn't fit, ask — don't bend the schema.

Workflow

1. Ensure connection  → load references/setup.md; verify with one read call
                        (e.g. get_dovetail_projects / search_workspace).
2. Discover           → list projects; with the user, pick the project(s) to sync.
3. Map                → load references/mapping.md; classify each Dovetail item:
                        Data entry → Transcripts/ ; Highlight → Signals/ ;
                        Insight → Evidence/ (or a Correlation candidate).
4. Plan               → show a table: Dovetail item → target file → type. Confirm.
5. Write              → create/update files with provenance frontmatter and
                        verbatim quotes; URL-encode links; add retrieved:/dates.
6. Wire               → link Signals to their Transcript + evidences:; surface new
                        Evidence to the relevant Persona; keep the graph consistent.
7. Freshness + gaps   → flag items older than 3 months (🕒); anything Dovetail
                        can't answer becomes a question in Research backlog.md.

What this skill never does

  • Never writes to Dovetail or changes Dovetail content.
  • Never commits API tokens or puts secrets in the repo.
  • Never overwrites a hand-written file without showing a diff and confirming.
  • Never turns a synthetic persona chat into a Dovetail record (that pipeline is one-way: Dovetail → us).

Re-sync

Files carry dovetail_id. On the next sync, match by that id: update the mapped file in place (content/quote/date), but keep our structure and any local edits the user flagged as kept. Never duplicate.

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.