Starter is public. Project HQ is gated.

Make project context easier to resume.

dovetell helps AI-assisted projects preserve recoverable intent: decisions, drift, handoffs, and repo-owned context that make the next session safer to resume.

repo-owned context starter
project-context/manifest.mdsource
project-context/tasks.mdactive
project-context/decisions.mdtrace
project-context/session-handoff.mdresume

The first proof surface is intentionally small: readable markdown near the work, with enough structure for a human or coding agent to resume safely.

Where to start

Start with the smallest context loop that helps.

Use the public resources to see the pattern, assess your team's current context habits, and choose the level of structure that fits the work in front of you.

Use now

nowBuy the Starter Context Kit or review the gated Project HQ and Team HQ paths.
nowTry the free sample set to see repo-owned context in practice.
nowTake the team assessment to find the weakest part of your AI context loop.
nowUse the framework to shape handoffs, decisions, and reusable prompts.

Being built

nextA clearer path from assessment results into the right product.
nextSaved assessment history for teams that want to track progress.
nextContext Inbox review for turning scattered notes into approved context.
nextWorkflow support that keeps human review in the loop.
Workflow

From scattered working memory to reusable team context.

The workflow is deliberately small: capture the things teams keep re-explaining, make them reviewable, and keep the durable version somewhere the team already owns.

01
Scattered inputs
Chats, PR comments, meeting notes, prompt fragments, and decisions in flight.
02
Proposed context
Useful material is shaped into a candidate update, not treated as truth by default.
03
Human review
Someone accountable accepts, rejects, edits, or parks the change.
04
Repo-owned context
The durable version lives as readable markdown near the work.
05
Agents reuse it
Future sessions start with less re-explaining and fewer fragile assumptions.

Start with the assessment.

Find the weakest part of your team's AI context loop, then route to the sample that fits the problem.

Take the assessment
Pain

The problem is not that teams lack tools. It is that context keeps escaping.

Repeated explanation

Every new chat starts with the same background. The best version of the explanation lives wherever someone last typed it.

Decision drift

Teams change direction during real work, but the durable record lags behind. Later, nobody can tell which decision still governs.

Handoff fragility

Work pauses, roles switch, or sessions reset. Without a compact handoff, useful context disappears at the exact moment it is needed.

Public proof

Choose a product without taking the assessment first.

The assessment is useful when you want a recommendation, but the kit paths are visible on their own: Starter, Project HQ, and Team HQ. The free sample set is there if you want to try the pattern first.

session-handoff.md
tasks.md

sample: inspect the repo-owned context loop

starter: preserve recoverable project intent

project hq: add governed docs and lineage

team hq: use docs as alignment controls

Writings

Give your team shared language for context debt.

Essays and field notes on context debt, reviewed team memory, and how teams keep AI-assisted work from starting cold.

writing/context-debt

thesis: AI makes stale context more expensive

pattern: scattered notes become agent-visible drift

move: reviewed context near the work

goal: fewer cold starts

Trust model

Built around repo-owned truth, not another private memory silo.

dovetell helps teams render, coordinate, review, and write back context without giving up ownership of the governed record.

source

Readable markdown

Context remains understandable without a proprietary app.

lineage

Traceable changes

Decisions and tasks keep source, status, and handoff history.

service

Bounded storage

Assessment and account state can live in Postgres without owning team truth.

guardrail

No magic claims

Useful governance beats dramatic promises about autonomous memory.

Early access

Get the next release note.

Join the list for Starter updates, saved assessment runs, and the next practical Context Inbox workflow.

You're on the list. We'll be in touch when the next release is ready.