boshu2__agentops/docs/index.md

title, description, hide

title	description	hide
AgentOps	SDLC control plane for coding agents. Keep the books, compile context, gate output, and compound knowledge so every session starts loaded, not cold.	navigation toc

AgentOps

The SDLC control plane for coding agents.

AgentOps keeps the books, compiles context, gates output, and compounds learning for agentic software development. Its substrate is .agents/: a wiki of markdown files in your repo, version-controlled with your code, that agents read, traverse, and contribute to.

AgentOps uses software-engineering practice people already understand — Agile/XP, BDD/Gherkin, DDD, hexagonal architecture, TDD, CI/CD, SRE, ADRs, provenance, and durable knowledge — then compiles those practices into dense, verifiable context for LLM agents under token scarcity. The internal lifecycle is the CDLC: context gets developed, tested, delivered, observed, and improved like any other software asset.

The only verifiable moat in this uncertain time is context. Models will get smarter, harnesses will commoditize, agents will get cheaper. Your accumulated context — the lessons learned about your individual problems, the patterns that worked, the decisions that survived review — is the one asset that compounds and doesn't get eaten by the next vendor release. That's what your company actually is.

AgentOps is the shovel. Start digging.

[:octicons-rocket-24: Install](#install){ .md-button .md-button--primary } [:octicons-play-24: See It Work](#see-it-work){ .md-button } [:octicons-mark-github-24: GitHub](https://github.com/boshu2/agentops){ .md-button }

---

The Problem

Every agent session starts cold. Same mistakes. Same rework. The landmine in auth.py, the two-hour timeout debug, the flag the reviewer always catches — none of it carries forward.

AgentOps solves this with four product layers:

Layer	What it does
Bookkeeping	Records what agents tried, changed, validated, and learned so the work leaves evidence
Context Compiler	Assembles the right context for the right phase — decay-ranked, token-budgeted, loaded at session start
Validation Gates	/pre-mortem, /vibe, and /council challenge plans and code before they ship
Knowledge Flywheel	Extracts learnings, scores them, and resurfaces them so the next session starts smarter

Session 1, your agent spends two hours debugging a timeout bug. Session 15, a new agent finds the lesson in seconds because the corpus kept it.

flowchart LR
    B[Bookkeeping] --> C[Context Compiler]
    C[Context Compiler] --> S[Session work]
    S --> G[Validation Gates]
    G --> F[Knowledge Flywheel]
    F --> B

All AgentOps state lives in local .agents/ — auditable, versionable, yours. Plain text you can grep, diff, review, and exclude from source control. No AgentOps-managed telemetry or hosted control plane; model runtimes, Git remotes, installers, and external tools are operator-selected dependencies. For constrained environments, see the Assurance Profile.

---

Install

Pick the runtime you use.

=== "Claude Code"

```bash
claude plugin marketplace add boshu2/agentops
claude plugin install agentops@agentops-marketplace
```

=== "Codex CLI (macOS / Linux / WSL)"

```bash
curl -fsSL https://raw.githubusercontent.com/boshu2/agentops/main/scripts/install-codex.sh | bash
```

=== "Codex CLI (Windows PowerShell)"

```powershell
irm https://raw.githubusercontent.com/boshu2/agentops/main/scripts/install-codex.ps1 | iex
```

=== "OpenCode"

```bash
curl -fsSL https://raw.githubusercontent.com/boshu2/agentops/main/scripts/install-opencode.sh | bash
```

Restart your agent after install, then type /quickstart in your agent chat.

The ao CLI is optional but recommended. It unlocks repo-native bookkeeping, retrieval, health checks, and terminal workflows.

=== "macOS"

```bash
brew tap boshu2/agentops https://github.com/boshu2/homebrew-agentops
brew install agentops
ao version
```

=== "Windows PowerShell"

```powershell
irm https://raw.githubusercontent.com/boshu2/agentops/main/scripts/install-ao.ps1 | iex
ao version
```

---

See It Work

Two commands. Real output.

Validate a PR with independent judges

> /council validate this PR

[council] 3 judges spawned independently
[judge-1] PASS — token bucket implementation correct
[judge-2] WARN — rate limiting missing on /login endpoint
[judge-3] PASS — Redis integration follows middleware pattern
Consensus: WARN — add rate limiting to /login before shipping

Full loop: research through post-mortem

> /rpi "add retry backoff to rate limiter"

[research]    Found 3 prior learnings on rate limiting
[plan]        2 issues, 1 wave
[pre-mortem]  Council validates the plan
[crank]       Executes the scoped work
[vibe]        Council validates the code
[post-mortem] Captures new learnings in .agents/
[flywheel]    Next session starts with better context

The point is not a bigger prompt. The point is a repo that remembers what was tried, what worked, what failed, and what should constrain the next run.

---

The headline skills

Every skill works alone. Compose flows for end-to-end cycles.

Skill	Use it when
/quickstart	You want the fastest setup check and next action
/council	You want independent judges to review a plan, PR, or decision
/research	You need codebase context and prior learnings before changing code
/pre-mortem	You want to pressure-test a plan before implementation
/implement	You want one scoped task built and validated
/rpi	You want discovery, build, validation, and bookkeeping in one flow
/vibe	You want a code-quality and risk review before shipping
/evolve	You want a goal-driven improvement loop with regression gates
/dream	You want overnight knowledge compounding that never mutates source code

!!! info "Full catalog" :octicons-book-24: All 173 skills — complete reference with source links and descriptions. :octicons-routes-24: Decision tree — "which skill do I need next?"

---

Unsupervised Cycles

Day: autonomous improvement. /evolve reads GOALS.md, fixes the worst fitness gap, runs regression gates, records each cycle.

> /evolve

[evolve] GOALS.md loaded
[cycle-1] Worst gap selected
[rpi]     Implements the fix
[gate]    Tests and quality checks pass
[learn]   Post-mortem feeds the flywheel

Night: knowledge compounding. /dream consolidates learnings, dedupes patterns, defragments the corpus. Purely bookkeeping work over .agents/ — source code stays untouched, /rpi never fires, no git operations.

> /dream start

[overnight] INGEST  harvest new artifacts
[overnight] REDUCE  dedup, defrag, close loops
[overnight] MEASURE corpus quality
[halted]    plateau reached

Morning report: .agents/overnight/<run-id>/summary.md

Run Dream overnight, then Evolve in the morning against a fresher corpus. Same model, smarter environment.

---

Next steps

1. Install — pick your runtime.
2. Seed your repo with ao quick-start (ao quickstart also works), then run /quickstart in your agent chat.
3. Choose a golden path: /rpi "a small goal" for a first validated change, /council validate this PR for review-only, or ao reconcile --json then bd ready to continue tracked work.

Read the lineage at 12factoragentops.com — DevOps applied to coding agents in twelve factors.

---

Explore

• :material-book-open: Newcomer Guide

  ---

  Repo orientation, mental model, and a fast path to becoming productive.

• :material-school: Levels L1–L5

  ---

  Progressive learning curriculum from single-session work to full autonomous orchestration.

• :material-console-line: CLI Reference

  ---

  Every ao command, flag, and exit code. Auto-generated on every build.

• :material-file-tree: Architecture

  ---

  System design: context compiler, validation gates, knowledge flywheel, RPI pipeline.

• :material-compare: Comparisons

  ---

  AgentOps vs Spec-Driven Development, Claude-Flow, Superpowers, Compound Engineer.

• :material-file-document-multiple: Contracts

  ---

  RPI run registry, finding registry, Dream runs, and every other inter-component contract.

---

Built by the AgentOps contributors. [Philosophy](philosophy.md) · [The Science](the-science.md) · [Strategic Direction](strategic-direction.md)