{"id":37341431,"url":"https://github.com/boshu2/agentops","last_synced_at":"2026-05-15T21:08:35.096Z","repository":{"id":331171253,"uuid":"1090518613","full_name":"boshu2/agentops","owner":"boshu2","description":"The operational layer for coding agents. Memory, validation, and feedback loops that compound between sessions.","archived":false,"fork":false,"pushed_at":"2026-05-01T07:05:30.000Z","size":56342,"stargazers_count":321,"open_issues_count":5,"forks_count":34,"subscribers_count":3,"default_branch":"main","last_synced_at":"2026-05-01T08:29:59.886Z","etag":null,"topics":["ai-agents","claude","claude-code","claude-code-plugins","claude-marketplace","claude-skills","codex","codex-plugin","cursor","devops","opencode-plugin","vibe-coding"],"latest_commit_sha":null,"homepage":"https://www.12factoragentops.com","language":"Go","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"other","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/boshu2.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","contributing":"docs/CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":"docs/CODE_OF_CONDUCT.md","threat_model":null,"audit":null,"citation":null,"codeowners":".github/CODEOWNERS","security":"docs/SECURITY.md","support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":"AGENTS.md","dco":null,"cla":null}},"created_at":"2025-11-05T19:18:56.000Z","updated_at":"2026-05-01T00:25:34.000Z","dependencies_parsed_at":"2026-02-07T03:04:45.904Z","dependency_job_id":null,"html_url":"https://github.com/boshu2/agentops","commit_stats":null,"previous_names":["boshu2/agentops"],"tags_count":137,"template":false,"template_full_name":null,"purl":"pkg:github/boshu2/agentops","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/boshu2%2Fagentops","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/boshu2%2Fagentops/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/boshu2%2Fagentops/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/boshu2%2Fagentops/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/boshu2","download_url":"https://codeload.github.com/boshu2/agentops/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/boshu2%2Fagentops/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":32596533,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-03T22:12:39.696Z","status":"online","status_checked_at":"2026-05-04T02:00:06.625Z","response_time":58,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["ai-agents","claude","claude-code","claude-code-plugins","claude-marketplace","claude-skills","codex","codex-plugin","cursor","devops","opencode-plugin","vibe-coding"],"created_at":"2026-01-16T03:51:24.411Z","updated_at":"2026-05-15T21:08:35.082Z","avatar_url":"https://github.com/boshu2.png","language":"Go","funding_links":[],"categories":["Workflow Infrastructure \u0026 Design"],"sub_categories":[],"readme":"\u003cdiv align=\"center\"\u003e\n\n# AgentOps\n\n[![GitHub stars](https://img.shields.io/github/stars/boshu2/agentops?style=social)](https://github.com/boshu2/agentops/stargazers)\n\n### The engineering operating system for agent teams.\n\nFrom agent opinions to engineering verdicts.\n\n\u003c!-- agentops:claim:AOP-CLAIM-README-FACTORY-CONTEXT --\u003e\n**AgentOps is a disciplined engineering layer for agentic software development.** It keeps the books for coding agents, then compiles that record into context for your software factory. It captures what agents tried, what worked, what failed, what was validated, and what should constrain the next run.\n\n*AgentOps is the shovel. Start digging.*\n\n\u003e AgentOps is not a coding harness. The labs are building those, and they will keep getting better. AgentOps sits on top of whichever harness you already use — Claude Code, Codex, Cursor, OpenCode — and turns your business, codebase, and team practices into a context library those agents mix and match from. Mix and match Claude, Codex, or any model at every phase. Lives in `.agents/` in your repo. Runs on your hardware. Evolves with the models.\n\n*AgentOps was used to develop AgentOps. As of 2026-05-04, this repo's `.agents/` directory contained ~1,842 learnings, ~186 patterns, ~80 planning rules, and ~3,867 cited decisions captured by the system on itself across thousands of phase transitions. Re-run anytime: `bash scripts/corpus-stats.sh`. Independent 3-judge audit (2026-05-06) confirmed parity with Anthropic Managed Agents on rubric authoring, separate-context grading, and iterate-until-pass.*\n\n\u003c/div\u003e\n\n---\n\n## Install\n\nPick the runtime you use.\n\n**Claude Code**\n\n```bash\nclaude plugin marketplace add boshu2/agentops\nclaude plugin install agentops@agentops-marketplace\n```\n\n**Codex CLI on macOS, Linux, or WSL**\n\n```bash\ncurl -fsSL https://raw.githubusercontent.com/boshu2/agentops/main/scripts/install-codex.sh | bash\n```\n\n**Codex CLI on Windows PowerShell**\n\n```powershell\nirm https://raw.githubusercontent.com/boshu2/agentops/main/scripts/install-codex.ps1 | iex\n```\n\n**OpenCode**\n\n```bash\ncurl -fsSL https://raw.githubusercontent.com/boshu2/agentops/main/scripts/install-opencode.sh | bash\n```\n\n**Other skills-compatible agents**\n\n```bash\nnpx skills@latest add boshu2/agentops --cursor -g\n```\n\nRestart your agent after install. Then type `/quickstart` in your agent chat.\n\nThe `ao` CLI is optional but recommended — repo-native bookkeeping, retrieval, health checks, and terminal workflows.\n\n**macOS**\n\n```bash\nbrew tap boshu2/agentops https://github.com/boshu2/homebrew-agentops\nbrew install agentops\nao version\n```\n\n**Windows PowerShell**\n\n```powershell\nirm https://raw.githubusercontent.com/boshu2/agentops/main/scripts/install-ao.ps1 | iex\nao version\n```\n\nYou can also install from [release binaries](https://github.com/boshu2/agentops/releases) or [build from source](cli/README.md). Troubleshooting: [docs/troubleshooting.md](docs/troubleshooting.md). Configuration: [docs/ENV-VARS.md](docs/ENV-VARS.md).\n\n---\n\n## See It Work\n\n**One command: per-phase model routing across Claude and Codex**\n\n```text\n$ ao rpi \"add rate limiting to /login\"\n[research/claude]    found 3 prior auth changes in .agents/decisions/\n[plan/claude]        proposed: token bucket, 5/min per IP, Redis-backed\n[pre-mortem/codex]   WARN: Redis unreachable case unhandled\n[implement/codex]    wrote middleware/ratelimit.go, 2 tests\n[validate/claude]    go test ./... PASS, gate: WARN — missing jitter\n[recorded]           .agents/runs/2026-05-07-rate-limit/\n```\n\nClaude does discovery, Codex implements, a fresh Claude validates, all in one loop with state preserved across boundaries. Nobody else does this.\n\nAnd when you want a second opinion before shipping: `/council --mixed`\n\n```text\n\u003e /council --mixed validate this PR\n\n[council] evidence packet sealed -\u003e 6 judges across Claude Code and Codex CLI\n[claude/judge-1] WARN - rate limiting missing on /login endpoint\n[claude/judge-2] PASS - Redis integration follows middleware pattern\n[codex/judge-1]  WARN - token bucket refill lacks jitter under burst\n[codex/judge-2]  PASS - backoff bounds match retry policy\nConsensus: WARN - fix /login rate limit and add refill jitter before shipping\nRecorded: .agents/council/\u003crun-id\u003e/verdict.md\n```\n\nThe 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.\n\n---\n\n\u003c!-- agentops:claim:AOP-CLAIM-README-FACTORY-CONTEXT --\u003e\n\n## What AgentOps Gives You\n\nFour layers. Each solves a different problem. All four compound.\n\n| Layer | Problem | What changes |\n|-------|---------|--------------|\n| **Bookkeeping** | Agents forget what they tried, why they changed course, and what evidence mattered | `.agents/` captures run packets, handoffs, findings, citations, decisions, verdicts, retros, and post-mortems. *The work leaves a trace.* |\n| **Context Compiler** | Every session starts from zero | `ao inject` delivers decay-ranked knowledge. `ao context assemble` builds phase-scoped packets. 75 skills load automatically via hooks. *Your agent starts loaded, not cold.* |\n| **Validation Gates** | Agents ship confident garbage | `/pre-mortem`, `/vibe`, `/council` — multi-model consensus validates plans before build and code before commit. Gates block, not advise. *Three fresh judges catch what one agent can't.* |\n| **Knowledge Flywheel** | Lessons disappear between sessions | `/forge` extracts learnings from the bookkeeping trail. `ao flywheel close-loop` scores and promotes. `/evolve` fixes the worst gap autonomously. `/dream` compounds overnight. *Session 15 starts with everything session 1 learned.* |\n\nAll state lives in local `.agents/` — plain text you can grep, diff, and review. No AgentOps-managed telemetry or hosted control plane. Runtime-neutral across Claude Code, Codex CLI, Cursor, and OpenCode.\n\n\u003c!-- agentops:claim:AOP-CLAIM-README-COMPETITIVE-MEMORY --\u003e\n\n### Why not just use Notion or Confluence?\n\n| Notion / Confluence | AgentOps `.agents/` |\n|---|---|\n| Written for humans; agents can't traverse it efficiently | LLM Wiki of Markdown — agents read it natively |\n| Lives in SaaS, not your repo | Lives in `.agents/` next to the code |\n| Not version-controlled with your code | Diffable, branchable, mergeable |\n| No decay ranking, no retrieval scoring | `ao inject` returns decay-ranked, token-budgeted packets |\n| No validation gates, no automated capture | Sessions write to it automatically; councils validate it |\n| Doesn't compound; you maintain it manually | Daemon defrags, evolves, and compounds it overnight |\n| Read-only artifact | Writes itself: agents that use it also produce it |\n\nMore: [docs/wiki-for-agents.md](docs/wiki-for-agents.md) · [docs/trust-factory.md](docs/trust-factory.md).\n\n---\n\n## Why DevOps?\n\nDevOps proved that disciplined systems around indeterministic workers produce reliable output. SRE proved it again with SLOs and error budgets. Kubernetes proved it for infrastructure with control loops. Coding agents are the next indeterministic worker class. Same playbook. New substrate.\n\nEvery primitive software engineering already gave us has a counterpart in the agent world:\n\n| Software Engineering | Coding-Agent World |\n|---|---|\n| Source code | Context (corpus, planning rules, learnings) |\n| SDLC | CDLC (Context Development Lifecycle) |\n| Libraries (Maven, npm, crates.io) | Context libraries (the `.agents/` corpus) |\n| Compilers | Context compilers (`ao compile` → wiki) |\n| Code review | Multi-model councils |\n| CI/CD | Validation gates (`/vibe`, `/pre-mortem`) |\n| Postmortems | Automated postmortems (`/post-mortem` → learnings) |\n| Runbooks | Skills + planning rules |\n| Software factories | Software factory daemon (`ao daemon`) |\n| Markdown / Git / Linux (open primitives) | LLM Wiki of Markdown |\n| Open-source corpus | Your private corpus (`.agents/` in your repo) |\n\nMajor engineering organizations are reorganizing around feeding their agents the right context — restructuring teams, building internal context platforms, hiring \"context engineering\" leads. AgentOps is that pattern for solo developers and small teams. Same playbook. Same asset class. Different scale.\n\nLLMs are engines. Context is fuel. You can't tune the engine — that's the model vendor's job. But you can engineer the fuel. Full CDLC treatment: [docs/cdlc.md](docs/cdlc.md).\n\n---\n\n## Quick Start\n\nInside a repo, use the path that matches what you are trying to do.\n\n\u003c!-- agentops:claim:AOP-CLAIM-README-FIRST-VALIDATED --\u003e\n\n| Path | Run | Done when |\n|------|-----|-----------|\n| **First repo setup** | `ao quick-start`, then `/quickstart` | AgentOps reports repo readiness and a next action |\n| **First validated change** | `/rpi \"a small goal\"` | Discovery, implementation, validation, and learning closeout leave evidence in `.agents/` |\n| **Review something now** | `/council validate this PR` or `/vibe recent` | You get a consolidated verdict and an evidence record in `.agents/` before shipping |\n\nNew project? Use the guided CLI seed first:\n\n```bash\nao quick-start     # Canonical\nao quickstart      # Stable alias\n```\n\nThat command applies the repeatable core seed: `.agents/`, `GOALS.md`, AgentOps instructions, starter knowledge, and readiness guidance. Use `/bootstrap` after that when you want the product/operations layer: `PRODUCT.md`, `README.md`, `PROGRAM.md`/`AUTODEV.md`, and optional hooks.\n\nAlready installed? Ask your agent for the next action:\n\n```text\n/quickstart\n```\n\nIf you installed the CLI, check your local setup:\n\n```bash\nao doctor\nao demo\n```\n\nThe demo path shows the 3.0 product loop: visible domain/practice packet,\nbounded task context, mixed Claude/Codex council verdict, tracked follow-up\nwork, and the optional scheduled compounding lane.\nFor the exact first-session command path, see\n[AgentOps 3.0 First-Value Path](docs/first-value-path.md). For copy, gist,\nand launch-post language, see the\n[AgentOps 3.0 Explainer Kit](docs/agentops-3-explainer-kit.md). For launch\nvideo planning, see the\n[AgentOps 3.0 YouTube Starter Series](docs/agentops-3-youtube-starter-series.md).\nFor the evidence loop behind PMF claims, see\n[AgentOps 3.0 PMF Evidence Loop](docs/agentops-3-pmf-evidence-loop.md).\n\nFull catalog: [docs/SKILLS.md](docs/SKILLS.md) · Unsure what to run? [Skill Router](docs/SKILL-ROUTER.md)\n\n---\n\n## Skills\n\nEvery skill works alone. Flows compose them when you want more structure.\n\n| Skill | Use it when |\n|-------|-------------|\n| `/quickstart` | You want the fastest setup check and next action |\n| `/council` | You want independent judges — optionally across Claude and Codex — to evaluate one evidence packet and return a consolidated verdict |\n| `/research` | You need codebase context and prior learnings before changing code |\n| `/pre-mortem` | You want to pressure-test a plan before implementation |\n| `/implement` | You want one scoped task built and validated |\n| `/rpi` | You want discovery, build, validation, and bookkeeping in one flow |\n| `/vibe` | You want a code-quality and risk review before shipping |\n| `/evolve` | You want a goal-driven improvement loop with regression gates |\n| `/dream` | You want overnight knowledge compounding that never mutates source code |\n\nFull reference: [docs/SKILLS.md](docs/SKILLS.md).\n\n---\n\n## The `ao` CLI\n\nThe `ao` CLI is the repo-native control plane behind the skills. It handles retrieval, health checks, compounding, goals, and terminal workflows.\n\n\u003c!-- agentops:claim:AOP-CLAIM-README-EVOLVE-AUTONOMOUS --\u003e\n\n```bash\nao quick-start                            # Set up AgentOps in a repo\nao doctor                                 # Check local health\nao demo                                   # See the council-first value path\nao search \"query\"                         # Search session history and local knowledge\nao lookup --query \"topic\"                 # Retrieve curated learnings and findings\nao context assemble                       # Build a task briefing\nao rpi phased \"fix auth startup\"          # Run the phased lifecycle from the terminal\nao evolve --max-cycles 1                  # Run one autonomous improvement cycle\nao overnight setup                        # Prepare private Dream runs\nao metrics health                         # Show flywheel health\n```\n\nFull reference: [CLI Commands](cli/docs/COMMANDS.md).\n\n---\n\n## Two ways to use it: hand agents and the software factory\n\n| Surface | When to use it | What it looks like | Operator role |\n|---------|---------------|-------------------|---------------|\n| **Hand agents** (skills surface) | Active work, exploration, high-stakes decisions, ambiguous scope | `/research`, `/plan`, `/pre-mortem`, `/council`, `/rpi` invoked from chat | Driving — agents respond, you steer |\n| **Software factory** (daemon) | Vetted, well-defined work; overnight compounding; bulk processing | `ao schedule` + `ao daemon` running dream / evolve / compile / defrag / forge unattended; mix-and-match councils per phase | Operator — you set cadence and quality bars; the factory runs |\n\n**Hand agents** — when you're driving. Skills (`/rpi`, `/council`, `/pre-mortem`, `/vibe`) work in chat. Different rigor levels available — light skills for exploratory work, full RPI loop for everything that should be tracked, council validation before shipping.\n\n\u003c!-- agentops:claim:AOP-CLAIM-README-AUTONOMOUS-FLYWHEEL --\u003e\n\n**Software factory** — when work is vetted and ready. The `ao daemon` runs scheduled jobs against your local subscription on your hardware. Mix and match models per phase: Claude for discovery, Codex for implementation, a fresh Claude for validation, an open-weights local model for overnight defrag. Run Dream overnight, then Evolve in the morning against a fresher corpus.\n\n→ [scheduling reference](docs/scheduling.md) · [example schedules](examples/schedules/).\n\n---\n\n## Docs\n\n| Topic | Where |\n|-------|-------|\n| Published site | [boshu2.github.io/agentops](https://boshu2.github.io/agentops/) |\n| Start navigating | [Docs index](docs/documentation-index.md) |\n| New contributor orientation | [Newcomer guide](docs/newcomer-guide.md) |\n| Working with `.agents/` | [Operator guide](docs/agents-operator-guide.md) |\n| Full skill catalog | [Skills](docs/SKILLS.md) |\n| CLI reference | [CLI commands](cli/docs/COMMANDS.md) |\n| Architecture | [Architecture](docs/ARCHITECTURE.md) |\n| FAQ | [FAQ](docs/FAQ.md) |\n\nAgentOps is built on the 12-factor doctrine — see [12factoragentops.com](https://12factoragentops.com).\n\n---\n\n## What if the labs ship this natively?\n\nThey will. Anthropic's Managed Agents is the first move; others will follow. That's fine — the value isn't in this tool. It's in the corpus you build with it.\n\nAgentOps is bridge infrastructure. Your `.agents/` directory is plain markdown in your repo. If a frontier vendor ships native equivalents in 12 months, your corpus carries forward. If we get acquired or change direction, your corpus is yours. If you outgrow the tool entirely, fork it, customize it, replace it — the corpus is what matters.\n\nOpen source forever. Built so you own the asset, not the tool.\n\n---\n\n## Contributing\n\nSee [docs/CONTRIBUTING.md](docs/CONTRIBUTING.md). Agent contributors should also read [AGENTS.md](AGENTS.md) and use `bd` for issue tracking.\n\n## License\n\nApache-2.0 · [Docs](docs/documentation-index.md) · [CLI Reference](cli/docs/COMMANDS.md)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fboshu2%2Fagentops","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fboshu2%2Fagentops","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fboshu2%2Fagentops/lists"}