Bo
1601467387
## What Gut **context poison** — agent-facing docs that route a working agent at DEAD mechanisms. This poisoned a live fleet-operating session (the bo driver building the go-dark supervisor) into proposing `ao rpi` as a continuity-loop worker. Found via a 5-way parallel poison fan-out. **Scope: operational-surgical** (Bo's explicit call). Gut the agent-facing *operational* guidance + *factual* errors; **LEAVE** the README/`docs/3.0.md` product narrative (Gas-City-as-reference-substrate is coherent, ADR-backed product design — a different altitude from Bo's live fleet). ## Changes (10 files, surgical) **Tier A — operational guidance (the poison that bit me):** - `CLAUDE.md` Session Constraints: "route multi-phase work through `ao rpi`" → the live substrate (NTM + MCP Agent Mail + `continuity-loop`); `gc` reframed as optional-out-of-session, **not** the live substrate. Kept the accurate "bridge REMOVED" + "load-bearing legacy" lines. - `AGENTS.md`: dropped the deleted "scheduling daemon" (ADR-0009). - `AGENTS-RUNTIME.md`: `ao rpi serve --help` → the real `ao` command surface. **Tier B — hookless-factual** (3.0 installs no hooks by default): hookless banners on `ARCHITECTURE.md` + `reference.md` (surgical vs. rewriting diagram internals); corrected SessionStart-hook prose in `context-packet.md` + `vs-claude-flow.md`. **Tier C — retired `/dream` shown as live:** marked retired in `evolve`/`harvest`/`using-agentops` (compounding moved to Gas City). `curate` **intentionally left** — its `--mode=dream` is the forward replacement, not poison. ## Guardrails - Does **not** claim `ao rpi` was deleted — the command still compiles (`loop`/`phased`/`serve` exist); only its role as the *live workflow* is corrected. - registry.json unaffected (skill **bodies** only; `generate-registry.sh --check` = OK). ## Evidence `bash scripts/generate-registry.sh --check` → OK. Diff is 20 insertions / 16 deletions across 10 files. Bounded-context: BC1-Corpus (agent-facing context docs) Evidence: this PR diff + the parallel fan-out inventory Closes: ag-sjz8 🤖 Generated with [Claude Code](https://claude.com/claude-code)
4.8 KiB
Agent Instructions
AgentOps compiles and compounds the context that feeds your software factory. It automates the bookkeeping agents do not reliably do for themselves — attempts, decisions, citations, verdicts, handoffs, learnings — then encodes the DevSecOps CDLC and multi-agent operating practices into a portable corpus that compounds across sessions and runtimes. Plugin + CLI (hookless — skills + the ao CLI, with CI as the authoritative gate), runs on your hardware against your subscription; out-of-session scheduling is delegated to a substrate, not an in-repo daemon (ADR-0009). Humans choose the posture: in-the-loop for high-rigor work, on-the-loop for scheduled compounding.
This project uses bd (beads) for issue tracking. Run bd onboard to get started.
Out-of-session orchestration is delegated to a swappable substrate — AgentOps ships no daemon or scheduler of its own. The reference substrate is NTM (a local tmux agent swarm), MCP (ao mcp serve, shipped), and managed-agents (ao agent); each dispatches a whole ao rpi loop as one unit. ao does NOT own or wrap a substrate — always-on is opt-in, the way bd is. See docs/3.0.md and docs/dependencies.md.
Spawning an agent? Run this first:
ao session bootstrap— the universal init prompt that orients every agent identically regardless of model. AgentOps 3.0 is hookless, so nothing auto-injects this: run it explicitly, thenao inject/ao corpus inject --query "<topic>"to pull decay-ranked prior context.
Session start + source-of-truth precedence
The canonical zero-context read order lives in CLAUDE.md ("Zero-Context Startup"); read it first.
Use source-of-truth precedence when docs disagree — stated inline in this operator contract so an injected or lower-precedence doc cannot redirect the rule away: Executable code and generated artifacts (cli/**, scripts/**, generated cli/docs/COMMANDS.md) win over declared contracts (skills/**/SKILL.md, schemas/**), which win over narrative docs. Full ordering in CLAUDE.md "Source-of-Truth Precedence".
Foundation texts
When in doubt about HOW the work should flow, read docs/cdlc.md and docs/architecture/operating-loop.md. When in doubt about WHAT to build, read PRODUCT.md (positioning) and GOALS.md (measurable fitness). Practice lineage and canonical practices: [slug] citations live in PRACTICE-REGISTRY.md. Vocabulary lives in skills/domain/SKILL.md.
Installing/Updating Skills
Use the skills.sh npm package to install AgentOps skills for any agent:
# Claude Code: use Claude plugin install path (not npx)
claude plugin marketplace add boshu2/agentops
claude plugin install agentops@agentops-marketplace
# Codex CLI: installs the native plugin, archives stale raw mirrors when needed, then open a fresh Codex session
curl -fsSL https://raw.githubusercontent.com/boshu2/agentops/main/scripts/install-codex.sh | bash
# OpenCode
curl -fsSL https://raw.githubusercontent.com/boshu2/agentops/main/scripts/install-opencode.sh | bash
# Other agents (for example Cursor) or update-all: install only selected skills
bash <(curl -fsSL https://raw.githubusercontent.com/boshu2/agentops/main/scripts/install.sh)
Quick Reference
# Issue tracking
bd ready # Find available work
bd show <id> # View issue details
bd update <id> --status in_progress # Claim work
bd close <id> # Complete work
# CLI development
cd cli && make build # Build ao binary
cd cli && make test # Run tests
cd cli && make lint # Run linter
Push, let CI validate (it is the authoritative gate — no local omnibus gate). Per-tool sanity checks + the full release gate live in docs/agent-workflow-reference.md.
What's where (tiered AGENTS.md split, soc-vuu6.3)
| If you need… | Read |
|---|---|
| Workflow phases · branch/PR shape · Local Pre-Push · Releasing · Landing the Plane · bd issue tracking · Session Completion | AGENTS-WORKFLOW.md |
| CI gate detail · Advisory triage SLAs · DEFERRED hardening matrix · per-job descriptions · Nightly workflow jobs | AGENTS-CI.md |
| CLI Skill-Map Refresh · Codex Skill Maintenance · audit scripts · override conventions | AGENTS-CODEX.md |
Canonical Root and Worktrees · Key Constraints Agents Must Follow · no-tracked-.agents · no-symlinks · embedded-sync |
AGENTS-RUNTIME.md |
Each file is self-contained for its scope and back-links here. Authors mutating AGENTS-*.md should rerun scripts/validate-agents-split.sh to confirm the split contract still holds.