Team reflection / 4Ls · Start–Stop–Continue · Sailboat

Retro

Leave the room with two clear changes the team can actually review.

2–3prioritized
actions
Maple Support / Cycle 12Fictional demonstration
84%42 completed
of 50 committed
Solid: 42 completedOutlined: 8 carried over
Liked

Buddy help

F1F2

2 notes · positive reported collaboration

Learned

Small checklists

F3F4

2 notes · missing details become visible

Lacked

The next owner

F5F6

2 notes · handoff responsibility unclear

Longed For

A queue cutoff

F7F8

2 notes · a shared decision rule

All four themes tie at two notes. Counts describe eight feedback notes, not eight unique respondents or a majority vote. The full-50-case goal was not achieved.

Retro: visual rendering of the saved fictional demonstration output
Fictional demonstration. A visual rendering of the actual saved Retro output from the source-guided Codex preparation run. Targets and proposals are not achieved business results. Select the image to open the full-resolution image.
01 / Read

Start with the cycle.

Bring the goal, commitments, results, raw feedback and previous actions.

02 / Understand

Find shared themes.

Group notes, show their frequency and preserve the evidence behind each interpretation.

03 / Decide

Keep the next steps small.

Choose 2–3 specific actions, each with an owner, deadline and success measure.

04 / Follow through

Return to last time.

Carry previous actions forward as Done, In Progress or Not Started, using the available evidence.

One useful meeting artifact.

A saved Markdown retrospective combines cycle performance, feedback themes, prioritized actions and carry-over status. Choose a format that fits the team; this does not create a shared whiteboard or run a meeting service.

Fictional saved example

42 / 50

Cases completed. That is 84%, with eight carried over. The result helps frame a constructive process discussion; it does not identify an employee to blame.

For a file-capable assistant.

The original English Skill is an instruction workflow for an AI assistant, not an installed team application. Supplied-text work needs no additional service API key. Model access or usage fees may apply; files require the host's file-reading tools.

It does not send feedback, schedule sessions, change a tracker or verify that a proposed improvement works. Review and approval stay with the team.

Source and local evidence.

Complete unchanged one-file subtree from phuryn/pm-skills, original MIT license and author attribution. The fictional Codex preparation run includes a normal case, an evidence boundary and local checks. Native-client installation, Claude execution and real team outcomes remain untested.