It reads past sessions.
Munin looks at your old Claude Code and Codex sessions and pulls out the parts that keep mattering.
It starts from the work you already did.
Features
What Munin actually does.
This page shows the loop from past sessions to a usable next step.
The full flow, step by step
Your sessions in. A handoff brief out. A check in the middle.
This is the loop: session history goes in, a short handoff comes out, and Munin checks that summary before trusting it.
Flow diagram. Three input moments - took a break, branch switch, interrupted debug - feed into the sessions journal. The journal passes through correction reasons and an auditable decision log, which become project memory. That memory emits a handoff brief and a next-step queue to the agent. A replay check sits below the memory as a safety check.
Read it top-to-bottom. The three moments at the top are where you lose context. They feed raw session data from Claude Code and Codex. Munin records correction reasons (bad-rule, wrong-scope, not-now, contradict) and writes every decision to an append-only log. That log becomes project memory - not an index you query, but the state the agent reads at session start. The memory emits a handoff brief and a next-step queue. On the right, a replay check decides whether the brief is safe to use. If the score drops, Munin falls back to deterministic packets instead of shipping stale memory.
It suggests fixes first.
Give Munin your priorities and it can point at the fix that matters most now, with a reason.
Less wandering. More useful first moves.
It keeps what still holds up.
Rules get stronger when they keep being right. Weak ones fade instead of hanging around forever.
You can see what looks solid and what does not.
Raw sessions become useful rules.
session stream - live
- [13:42:11] claude-code - session_b9f2 - started
- [13:42:18] corrected
prisma.migration->prisma.db.push - [13:42:44] verified via
pnpm test - [13:43:02] codex - session_c41a - redirect +3
- [13:43:19] confirmed
pnpm run lintclean - [13:43:48] reconciled 6 candidates against past sessions
promoted rules
-
Use
prisma db pushfor local schema syncfresh 0.94 conf 0.88 stab 0.71 -
Run
pnpm testbefore commit on sitesortedfresh 0.81 conf 0.92 stab 0.85 -
Legacy webpack override forgottenfresh 0.12 conf 0.38 stab 0.22
Workspace organisation per project
Munin keeps separate memory per project, so the agent is less likely to drag old context into the wrong repo.
The right project gets the right memory.
It shows repeated mistakes.
Munin shows what each agent keeps getting wrong so you can fix the pattern instead of correcting it by hand.
Shared project memory. Different fixes per agent.
It checks itself.
Munin replays old sessions to see whether the handoff brief still works. If it does not, it falls back.
No quiet failures.
Morning proactivity
munin proactivity runs the morning check, queues a fix, and can launch the right agent session with that fix already framed.
Agents suggest first instead of waiting.
One dashboard per project. Repeated mistakes named, not guessed.
workspaces
- sitesorted 3 open loops 12 active memories
- orchestrator-v2 1 open loop 7 active memories
- peakflow idle 4 active memories
- textbee 2 open loops 9 active memories
repeated-mistakes report
sitesorted - last 14d
-
codex ignored prettier config x7-> inject
.prettierrcpath in agent preamble -
claude-code re-invented webhook header x3-> pin
x-munin-sigrule to memory -
ide skipped type-check on save x4-> add
tsc --noEmitto pre-commit gate
Every next step arrives with evidence.
top next step - 2026-04-14
confidence 0.82
Land retry-window fix before Thursday freeze.
evidence chain
- sessions4 referenced in last 6 days
- corrections2 recorded against current retry loop
- ticketMUN-214 open, owner @bridget
- expected effectremoves 1 open loop, closes 1 active memory