RPI Workflow
You have access to workflow skills for structured development.
The RPI Workflow
Research β Plan β Implement β Validate
β β
βββββ Knowledge Flywheel βββββ
Research Phase
/research <topic>
ao search "<query>"
ao search "<query>" --cite retrieved
ao lookup <id>
ao lookup --query "x"
Output: .agents/research/<topic>.md
Plan Phase
/pre-mortem <spec>
/plan <goal>
Output: Beads issues with dependencies
Implement Phase
/implement <issue>
/crank <epic>
/swarm
Output: Code changes, tests, documentation
Validate Phase
/vibe [target]
/post-mortem
/retro
Output: .agents/learnings/, .agents/patterns/
Release Phase
/release [version]
/release --check
/release --dry-run
Output: Updated CHANGELOG.md, version bumps, git tag, docs/releases/
Phase-to-Skill Mapping
| Phase |
Primary Skill |
Supporting Skills |
| Discovery |
/discovery |
/brainstorm, /research, /plan, /pre-mortem |
| Implement |
/crank |
/implement (single issue), /swarm (parallel execution) |
| Validate |
/validation |
/vibe, /post-mortem, /retro, /forge |
| Release |
/release |
β |
Choosing the skill:
- Use
/implement for single issue execution. Now defaults to TDD-first β writes failing tests before implementing. Skip with --no-tdd.
- Use
/crank for autonomous epic execution (loops waves via swarm until done). Auto-generates file-ownership maps to prevent worker conflicts.
- Use
/swarm directly for parallel execution without beads (TaskList only).
- Use
/discovery for the discovery phase only (brainstorm β search β research β plan β pre-mortem).
- Use
/validation for the validation phase only (vibe β post-mortem β retro β forge).
- Use
/rpi for full lifecycle β delegates to /discovery β /crank β /validation.
- Use
/ratchet to gate/record progress through RPI.
Available Skills
Start Here (12 starters)
These are the skills every user needs first. Everything else is available when you need it.
| Skill |
Purpose |
/quickstart |
Guided onboarding β run this first |
/bootstrap |
One-command full AgentOps setup β fills gaps only |
/research |
Deep codebase exploration |
/council |
Multi-model consensus review + finding auto-extraction |
/vibe |
Code validation (classification + suppression + domain checklists) |
/rpi |
Full RPI lifecycle orchestrator (/discovery β /crank β /validation) |
/implement |
Execute single issue |
/retro --quick |
Quick-capture a single learning into the flywheel |
/status |
Single-screen dashboard of current work and suggested next action |
/goals |
Maintain GOALS.yaml fitness specification |
/push |
Atomic test-commit-push workflow |
/flywheel |
Knowledge flywheel health monitoring (ΟΓΟ > Ξ΄/100) |
Advanced Skills (when you need them)
| Skill |
Purpose |
/compile |
Active knowledge intelligence β Mine β Grow β Defrag cycle |
/harvest |
Cross-rig knowledge consolidation β sweep, dedup, promote to global hub |
/knowledge-activation |
Operationalize a mature .agents corpus into beliefs, playbooks, briefings, and gap surfaces |
/brainstorm |
Structured idea exploration before planning |
/discovery |
Full discovery phase orchestrator (brainstorm β search β research β plan β pre-mortem) |
/plan |
Epic decomposition into issues |
/design |
Product validation gate β goal alignment, persona fit, competitive differentiation |
/pre-mortem |
Failure simulation (error/rescue, scope modes, temporal, predictions) |
/post-mortem |
Validation + streak tracking + prediction accuracy + retro history |
/bug-hunt |
Root cause analysis |
/release |
Pre-flight, changelog, version bumps, tag |
/crank |
Autonomous epic loop (uses swarm for each wave) |
/swarm |
Fresh-context parallel execution (Ralph pattern) |
/evolve |
Goal-driven fitness-scored improvement loop |
/doc |
Documentation generation |
/retro |
Quick-capture a learning (full retro β /post-mortem) |
/validation |
Full validation phase orchestrator (vibe β post-mortem β retro β forge) |
/ratchet |
Brownian Ratchet progress gates for RPI workflow |
/forge |
Mine transcripts for knowledge β decisions, learnings, patterns |
/readme |
Generate gold-standard README for any project |
/security |
Continuous repository security scanning and release gating |
/security-suite |
Binary and prompt-surface security suite β static analysis, dynamic tracing, offline redteam, policy gating |
/test |
Test generation, coverage analysis, and TDD workflow |
/red-team |
Persona-based adversarial validation β probe docs and skills from constrained user perspectives |
/review |
Review incoming PRs, agent output, or diffs β SCORED checklist |
/refactor |
Safe, verified refactoring with regression testing at each step |
/deps |
Dependency audit, update, vulnerability scanning, and license compliance |
/perf |
Performance profiling, benchmarking, regression detection, and optimization |
/scaffold |
Project scaffolding, component generation, and boilerplate setup |
/scenario |
Author and manage holdout scenarios for behavioral validation |
Expert Skills (specialized workflows)
| Skill |
Purpose |
/grafana-platform-dashboard |
Build Grafana platform dashboards from templates/contracts |
/codex-team |
Parallel Codex agent execution |
/openai-docs |
Official OpenAI docs lookup with citations |
/oss-docs |
OSS documentation scaffold and audit |
/reverse-engineer-rpi |
Reverse-engineer a product into feature catalog and specs |
/pr-research |
Upstream repository research before contribution |
/pr-plan |
External contribution planning |
/pr-implement |
Fork-based PR implementation |
/pr-validate |
PR-specific validation and isolation checks |
/pr-prep |
PR preparation and structured body generation |
/pr-retro |
Learn from PR outcomes |
/complexity |
Code complexity analysis |
/product |
Interactive PRODUCT.md generation |
/handoff |
Session handoff for continuation |
/recover |
Post-compaction context recovery |
/trace |
Trace design decisions through history |
/provenance |
Trace artifact lineage to sources |
/beads |
Issue tracking operations |
/heal-skill |
Detect and fix skill hygiene issues |
/converter |
Convert skills to Codex/Cursor formats |
/update |
Reinstall all AgentOps skills from latest source |
Knowledge Flywheel
Every /post-mortem feeds back to /research:
- Learnings extracted β
.agents/learnings/
- Patterns discovered β
.agents/patterns/
- Research enriched β Future sessions benefit
Runtime Modes
AgentOps has four runtime modes. Do not assume hook automation exists everywhere.
| Mode |
When it applies |
Start path |
Closeout path |
Guarantees |
gc |
Gas City (gc) binary available and city.toml present |
gc controller manages sessions; ao rpi auto-selects gc executor |
gc event bus captures phase/gate/failure/metric events |
Default when gc is available. Phase execution via gc sessions, events via gc event bus, agent health via gc health patrol |
hook-capable |
Claude/OpenCode with lifecycle hooks installed (no gc) |
Runtime hook or ao inject / ao lookup |
Runtime hook or ao forge transcript + ao flywheel close-loop |
Automatic startup/context injection and session-end maintenance when hooks are installed |
codex-hookless-fallback |
Codex Desktop / Codex CLI without hook surfaces |
ao codex start |
ao codex stop |
Explicit startup context, citation tracking, transcript fallback, and close-loop metrics without hooks |
manual |
No hooks and no Codex-native runtime detection |
ao inject / ao lookup |
ao forge transcript + ao flywheel close-loop |
Works everywhere, but lifecycle actions are operator-driven |
Issue Tracking
This workflow uses beads for git-native issue tracking:
bd ready
bd show <id>
bd close <id>
bd vc status
Examples
Startup Context Loading
Hook-capable runtimes
session-start.sh (or equivalent) can run at session start.
- In
manual mode, MEMORY.md is auto-loaded and the hook points to on-demand retrieval (ao search, ao lookup).
- In
lean mode, the hook extracts pending knowledge and injects prior learnings with a reduced token budget.
- This skill can be injected automatically into session context.
Codex hookless fallback
- Run
ao codex start.
- AgentOps inspects
.agents/, runs safe close-loop maintenance, syncs MEMORY.md, and writes .agents/ao/codex/startup-context.md.
- Surfaced learnings, patterns, and findings are cited as
retrieved.
- Use
ao lookup for automatic citations during work, or ao search --cite retrieved|reference|applied when a search result is actually adopted.
- End the session with
ao codex stop, then verify loop health with ao codex status.
Result: The agent gets the RPI workflow, prior context, and a citation path in both modes. The difference is whether lifecycle work is hook-driven or command-driven.
Workflow Reference During Planning
User says: "How should I approach this feature?"
What happens:
- Agent references this skill's RPI workflow section
- Agent recommends Research β Plan β Implement β Validate phases
- Agent suggests
/research for codebase exploration, /plan for decomposition
- Agent explains
/pre-mortem for failure simulation before implementation
- User follows recommended workflow with agent guidance
Result: Agent provides structured workflow guidance based on this meta-skill, avoiding ad-hoc approaches.
Troubleshooting
| Problem |
Cause |
Solution |
| Skill not auto-loaded |
Hook runtime unavailable or startup path not run |
Hook-capable runtimes: verify hooks/session-start.sh exists and is enabled. Codex: run ao codex start explicitly |
| Outdated skill catalog |
This file not synced with actual skills/ directory |
Update skill list in this file after adding/removing skills |
| Wrong skill suggested |
Natural language trigger ambiguous |
User explicitly calls skill with /skill-name syntax |
| Workflow unclear |
RPI phases not well-documented here |
Read full workflow guide in README.md or docs/ARCHITECTURE.md |