diegosouzapw/awesome-omni-skill / skills/ai-agents/agent-ops-retrospective
agent-ops-retrospective
VerifiedScan the current chat session for durable learnings (clarifications, corrections, decisions, pitfalls) and update .agent/memory.md. Use after critical review and before concluding work.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 3.4 KB |
| metadata.json | config | 1.5 KB |
SKILL.md
View on GitHub---
name: agent-ops-retrospective
description: "Scan the current chat session for durable learnings (clarifications, corrections, decisions, pitfalls) and update .agent/memory.md. Use after critical review and before concluding work."
license: MIT
compatibility: [opencode, claude, cursor]
metadata:
category: core
related: [agent-ops-state, agent-ops-tasks]
---
# Retrospective / Learning workflow (mandatory)
## Goal
Ensure durable, reusable insights from this chat session are captured in `.agent/memory.md` so future sessions can resume without re-discovering them.
## Inputs
- The current chat session transcript (user + assistant messages in this session)
- `.agent/constitution.md`, `.agent/memory.md`, `.agent/focus.md`, `.agent/issues/`
## Extraction rules (strict)
Only capture *durable* items:
- confirmed workflow rules
- stable project conventions (not one-off)
- confirmed commands (but commands/boundaries belong in constitution, not memory)
- user preferences that affect future work on this repo
- corrections to previous misunderstandings
- pitfalls/gotchas discovered during implementation
Do NOT capture:
- transient task state (belongs in focus)
- speculative ideas not adopted
- secrets/tokens
- personal/sensitive data
## Placement rules (strict)
- Project-specific commands/boundaries/constraints → `.agent/constitution.md`
- Durable workflow learnings and recurring conventions → `.agent/memory.md`
- Current session status → `.agent/focus.md`
- Follow-ups and approvals needed → `.agent/issues/`
## Procedure
1) Read: constitution/memory/focus/tasks.
2) Scan the chat session for:
- explicit corrections ("No, do X instead of Y")
- newly confirmed commands or tools
- newly confirmed constraints ("never refactor", "only write docs in …")
- repeated misunderstandings (add a "pitfall to avoid")
- preferences expressed by the user
3) Update `.agent/memory.md`:
- append a dated subsection: `## Retrospective YYYY-MM-DD`
- add short, atomic bullets phrased as "Do/Don't/Prefer"
- avoid duplication with constitution (link to constitution section if needed)
4) Update focus if the retrospective reveals unresolved items.
5) Invoke `agent-ops-tasks` discovery if actionable items found.
6) Do not declare completion unless retrospective has been run.
## Issue Discovery During Retrospective
**After scanning session, invoke `agent-ops-tasks` discovery for actionable items:**
1) **Collect actionable learnings:**
- "We should add tests for X" → `TEST` issue
- "This pattern is confusing, needs docs" → `DOCS` issue
- "Found a bug but didn't fix it" → `BUG` issue
- "This could be optimized later" → `PERF` issue
- "Technical debt noticed" → `CHORE` or `REFAC` issue
- "Security concern noted" → `SEC` issue
2) **Present to user:**
```
📋 Retrospective identified {N} actionable items:
Medium:
- [DOCS] Document the retry mechanism in PaymentService
- [TEST] Add integration tests for new OAuth flow
Low:
- [CHORE] Clean up commented-out code in UserController
- [PERF] Consider caching user preferences (noted during implementation)
Create issues for these? [A]ll / [S]elect / [N]one
```
3) **After creating issues:**
- Mark in memory.md: "Created {ISSUE-IDs} for follow-up"
- These become part of the project backlog
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |