Skip to main content

From session to board. Nothing important lost.

Connect Jentrix to your Claude Code or Codex session with one prompt. The plan, the decisions, the outputs, the issues and the gaps land on the task card — typed, on a board you can actually review.

one prompt to connect · free for one person · bring your own agent

Works withClaude Codeby AnthropicCodexby OpenAI
claude codecodexSESSIONmake the board load faster and prove itJentrix UI/UXBOARDBACKLOGIN REVIEWDONEJEN-73JEN-75JEN-76JEN-84182.4M tok · 42m · $206JEN-85JEN-2acceptedJEN-7acceptedPLANDECISIONOUTPUTISSUEGAP

A session is a wall of text.

Thousands of lines, tool calls and diffs scroll past while the agent works. The plan, the decisions, what it found broken and what it left undone are in there — and nobody is going to read it twice.

182.4Mtokens, read and written
42minutes of agent wall-clock
$206of model spend — your provider's bill, never ours
8artifacts Jentrix kept — 2 inputs · 1 plan · 2 outputs · 1 gap · 1 learning · 1 record

JEN-84 · one real session, as the board records it

Connect once. It collects what matters.

One prompt, pasted into Claude Code or Codex, installs the CLI and binds the session to a card on your board. From then on the agent files its record as typed artifacts — the opening prompt by itself, the rest as it works.

  1. input
  2. plan
  3. decision
  4. output
  5. issue
  6. gap
  7. learning
  8. record

13 push kinds · 8 review categories · typed, attributed, redacted

A Jentrix task panel: seven artifacts grouped as input, output, issue and gap, each attributed to the person and the agent session that produced it, and a review section showing the task accepted with every artifact reviewed

Kept at the moments a session loses things.

When it starts, when its context compacts, when it ends — the record is written there, not reconstructed afterwards.

at align · at compaction · at close

  • The opening prompt files itself the moment the session is aligned.
  • Before Claude Code compacts its context, a snapshot of the transcript lands first.
  • Every run closes with a RUN_SUMMARY: what happened, what is missing, whether it is review-ready.
  • The full transcript stays one click away on the session page.
  • Tokens, minutes and dollars ride on the card.
  • Secrets are redacted at ingestion; an unknown stays unknown.
The Jentrix My Work page: twenty tasks awaiting your review, grouped by board, each row carrying its artifact categories and keyboard shortcuts to move, open, select and accept

Review it your way.

Walk the record yourself, or have an agent walk it first. Either way the verdict is on the task — and accepting is a human control.

You review

  • One queue of everything awaiting you, across boards.
  • Walk the eight categories in order — input to record.
  • Tick what you checked; accept or return with a reason.
  • The verdict is recorded: who, when, on which artifacts.

An agent reviews — you decide

  • Run the review command in Claude Code or Codex when you want a second pair of eyes.
  • It walks the same eight categories over MCP and checks the readiness contract itself.
  • Its verdict lands as one comment on the task — it never moves the card.
  • Accept or return stays yours.

Try it now. It's free.

Every feature, free for one person. Bring your own agent — Claude Code or Codex — and we never charge for tokens.

  1. 01Request access; once approved, sign in with Google or an email link.
  2. 02Paste one prompt into Claude Code or Codex — it installs the CLI, signs you in and connects your folder.
  3. 03Work as usual. Align the session to a card; its record lands there as it goes.

no card · no token charges

A Jentrix board in light mode: kanban columns whose cards carry live token counts, minutes and dollars, and chips counting the artifacts each session produced

Agents do the work. You keep the record — and the decision.

Bring your own agent and your own keys. Jentrix keeps the record — free for one person.