SourceA · Chat Unify
Chat Unify
Prove every agent action — paste replies, verify with PASS or BLOCK, audit claims to disk receipts, and unify sessions into one brief. Your working desk for governed agent work.
Connecting…
Paste · verify · unify
Drop in any chat extract or agent reply. Verify it, audit claims to disk, forge a mission, or save to your session library — receipts stay local.
Checking status…
Save session
Library
Output · select item
No output yet.
Unified brief
Save sessions, then tap Unify sessions to build your brief.
Analysis findings
Cross-session contradictions and risks flagged by local rules — these are review signals from your saved chats, not defects in this app.
Advanced tools · structure · export · n8n
Getting started
Welcome to Chat Unify
Your governed SourceA workspace — paste agent replies, verify with PASS or BLOCK, audit claims to receipts, and unify sessions. Everything below is inside this app.
Recommended first steps
Pick your path
Three common starting points — all inside Chat Unify, no extra setup.
All twelve machines
Every machine in Chat Unify — tap to open. Nothing here is a separate product.
Help & updates
Checking for updates…
Ready to paste?
Verify & Act
7 STEPSPaste one agent reply. Each step runs left to right — plain language, reasoning, proof on disk, then a gated action you can trust.
Pro options · auto-run · Cursor send
-
1Language PLAIN
—
-
2Reasoning THINK
—
-
3Proof VERIFY
—
-
4Action ACT
—
-
5AI advisor ADVISE
—
-
6AI critic CRITIC
—
-
7Close loop CLOSE
—
Audit Trail
REVERSE · TRACE 7 STEPSPaste raw AI output. We parse claims, trace them to disk, and flag what cannot be verified — so you see the receipt, not blind trust.
-
1Parser PARSE
—
-
2Classifier TYPE
—
-
3Consistency CHECK
—
-
4Source TRACE
—
-
5Simplifier PLAIN
—
-
6Red flags FLAG
—
-
7Report OUT
—
Record a founder decision — approve scope, block work, or set permissions before agents proceed.
Queue and dispatch agent tasks — every outcome writes a receipt you can verify.
Skills, upgrade checklists, and the experience log for this app — for teams running governed agents.
Prompt Forge
4 STEPS OUTBOUND · #3Turn founder language into one bounded Cursor mission — policy check first, optional AI polish, receipt on disk. Email: forge@sourcea.app
-
1Lint POLICY
—
-
2Extract FACTS
—
-
3Compile MISSION
—
-
4Emit DISK
—
Proof Pack
5 STEPS DELIVER PROOFTurn a verified green run into a sealed, replayable proof you can share — the receipt, not the recipe. Truth gate must PASS before seal.
-
1Collect PULL
—
-
2Verify GATE
—
-
3Seal HASH
—
-
4Package VIEWS
—
-
5Emit DISK
—
Vocabulary Intelligence
6 STEPS FOUNDER WORDSScan any source for retired motor words — repo, live URL, paste, or local file. Classify by tier · suggest Cloud Forge Run + Auto Runtime. Optional LLM review when OpenRouter is wired.
-
1Goal CAMPAIGN
—
-
2Scan DISK
—
-
3Classify TIER
—
-
4Suggest WORD
—
-
5Review QUEUE
—
-
6Emit RECEIPT
—
Connect · integration hub
Wire your stack once. Run governed agents everywhere.
Link Cursor, cloud workers, and automation — live status probes with honest PASS or OFF, not green theater.
Loading integrations…
Inbound webhooks · automation triggers
POST JSON from n8n, Zapier, CI, or your own worker. Events route to Prompt Forge, Verify, or Proof Pack.