openup-retrospective

Generate iteration retrospective with feedback and action items

6 stars

Best use case

openup-retrospective is best used when you need a repeatable AI agent workflow instead of a one-off prompt.

Generate iteration retrospective with feedback and action items

Teams using openup-retrospective should expect a more consistent output, faster repeated execution, less prompt rewriting.

When to use this skill

  • You want a reusable workflow that can be run more than once with consistent structure.

When not to use this skill

  • You only need a quick one-off answer and do not need a reusable workflow.
  • You cannot install or maintain the underlying files, dependencies, or repository context.

Installation

Claude Code / Cursor / Codex

$curl -o ~/.claude/skills/openup-retrospective/SKILL.md --create-dirs "https://raw.githubusercontent.com/GermanDZ/open-up-for-ai-agents/main/docs-eng-process/.claude-templates/skills/openup-retrospective/SKILL.md"

Manual Installation

  1. Download SKILL.md from GitHub
  2. Place it in .claude/skills/openup-retrospective/SKILL.md inside your project
  3. Restart your AI agent — it will auto-discover the skill

How openup-retrospective Compares

Feature / Agentopenup-retrospectiveStandard Approach
Platform SupportNot specifiedLimited / Varies
Context Awareness High Baseline
Installation ComplexityUnknownN/A

Frequently Asked Questions

What does this skill do?

Generate iteration retrospective with feedback and action items

Where can I find the source code?

You can find the source code on GitHub using the link provided at the top of the page.

SKILL.md Source

# Retrospective

Generate an iteration retrospective capturing what went well, what to improve, and action items.

## Process

### 1. Determine Iteration

If `$ARGUMENTS[iteration_number]` is provided, use it. Otherwise read `docs/project-status.md` for the current iteration number.

### 2. Read Project Context

Read `docs/project-status.md` for: iteration goal, dates, team members, overall status.

### 3. Analyze Completed Tasks

Read `docs/roadmap.md` to identify: tasks planned, completed, not completed, and added during iteration. Note complexity, challenges, and successes for each.

### 4. Gather Feedback

Review these sources for patterns and issues:
- `docs/agent-logs/` - Agent run logs
- `docs/risk-list.md` - Risks emerged or mitigated
- `docs/roadmap.md` - Velocity (completed vs planned), blocked items
- Git commit messages

### 4b. Measure Read-Back (success measures whose date has passed)

This is the step that closes the loop between per-feature success measures
(task-spec `## Success Measures`, rubric criterion 12) and value
prioritization — without it, measures are write-only and the roadmap ordering
stays opinion-based.

1. Scan archived change folders (`docs/changes/archive/T-NNN/design.md`) for
   recorded success-measure grades + read-back dates (written by
   `/openup-complete-task` step 1b). Skip `n/a` entries.
2. For each entry whose **read-back date has passed** and has no recorded
   outcome yet: read the instrumentation named in the expectation (the event /
   metric / query) and record **actual vs expected** — including "instrumentation
   exists but nobody can produce the number" (that is a finding, not a skip).
3. Write the results into the retrospective document's **Measure Read-Back**
   section (see step 6): expectation, actual, verdict (met / missed / can't
   tell), one-line interpretation.
4. **Hand the section to the product-manager role** (`.claude/teammates/product-manager.md`):
   it consumes these verdicts to re-rank pending roadmap entries, updating each
   moved entry's `Value` rationale to cite the evidence ("UC-12's measure
   missed by 80% — demoting the follow-on entries below the X work"). The
   re-rank is the product-manager's call; the retrospective only delivers the
   evidence.
5. Note read-backs that come due **before** the next expected retrospective as
   action items with owners, so they aren't silently skipped.

### 5. Collect Metrics (if `$ARGUMENTS[include_metrics] == "true"`)

```bash
# Commits in iteration period
git log --oneline --since="$start_date" --until="$end_date" | wc -l

# Lines changed
git diff --stat trunk...HEAD

# Active contributors
git shortlog -sn --since="$start_date" --until="$end_date"
```

Task metrics: tasks planned, tasks completed, completion rate (completed / planned * 100%).

### 6. Create Retrospective Document

Create `docs/iteration-retrospectives/iteration-{n}-retrospective.md` with sections:
- **Iteration Overview**: number, date range, goal, participants
- **Summary**: overall assessment, key achievements, major challenges
- **What Went Well**: process, technical, collaboration successes
- **What to Improve**: process issues, technical challenges, gaps
- **Measure Read-Back**: for each success measure due (step 4b) — expectation, actual, verdict (met / missed / can't tell), interpretation; plus the product-manager's resulting re-rank decisions (entries moved + updated `Value` rationale), or "no re-rank — evidence supports current order"
- **Action Items**: specific action, owner, due date, priority for each improvement
- **Metrics** (if included): task completion stats, git stats
- **Next Iteration Considerations**: carry forward, changes, risks to monitor

### 7. Update Project Status

In `docs/project-status.md`: add link to retrospective, note ongoing action items, update iteration status.

### 8. Reset the Retro-Cadence Counter (T-011)

Running this retrospective satisfies the cadence, so reset the durable counter. This zeroes
`.openup/retro.json` and clears `gates.retro_due` in any live `.openup/state.json`:

```bash
python3 scripts/openup-state.py retro reset
```

After this, `/openup-start-iteration` will permit `full`-track starts again until 5 more
tasks complete. See [state-file.md](../../../../docs-eng-process/state-file.md).

## Output

Returns: retrospective document path, counts of what went well / what to improve / action items, overall iteration rating, key metrics (if included).

## See Also

- [openup-start-iteration](../start-iteration/SKILL.md) - Start next iteration
- [openup-complete-task](../complete-task/SKILL.md) - Complete iteration tasks
- [openup-assess-completeness](../assess-completeness/SKILL.md) - Assess iteration completeness before retrospective
- [openup-create-iteration-plan](../openup-artifacts/create-iteration-plan/SKILL.md) - Plan next iteration based on retrospective

Related Skills

openup-transition

6
from GermanDZ/open-up-for-ai-agents

Initialize and manage Transition phase activities - deploy to users

openup-tdd-workflow

6
from GermanDZ/open-up-for-ai-agents

Guide Test-Driven Development cycle adapted for AI agents with a pragmatic approach

openup-sync-spec

6
from GermanDZ/open-up-for-ai-agents

Back-propagate pure refactors to stale artifacts; classify the diff, refuse behaviour-changes, propose targeted edits for approval (read-only by default)

openup-start-iteration

6
from GermanDZ/open-up-for-ai-agents

Begin a new OpenUP iteration with proper phase context and task selection

openup-shared-vision

6
from GermanDZ/open-up-for-ai-agents

Create shared technical vision for team alignment

openup-request-input

6
from GermanDZ/open-up-for-ai-agents

Create an input request document for asynchronous stakeholder communication

openup-readiness

6
from GermanDZ/open-up-for-ai-agents

Compute the change-folder dependency DAG and print READY/BLOCKED/collision report for PM intake

openup-quick-task

6
from GermanDZ/open-up-for-ai-agents

Fast iteration mode for small changes - simplified workflow with minimal overhead

openup-plan-feature

6
from GermanDZ/open-up-for-ai-agents

Generate iteration plan and roadmap entry for a feature idea

openup-phase-review

6
from GermanDZ/open-up-for-ai-agents

Check phase completion criteria and prepare for phase review

openup-orchestrate

6
from GermanDZ/open-up-for-ai-agents

Run a full orchestrated iteration — PM decomposes the goal, delegates to specialist roles, collects outputs, and synthesizes results

openup-next

6
from GermanDZ/open-up-for-ai-agents

Run ONE OpenUP delivery cycle — read the derived board, claim the top READY lane, execute it, tick its progress, and exit through a legal exit. The sequential continue-loop.