# Openclaw Optimizer

> Use when: you want to optimize an OpenClaw setup (v2026.2.23+) — cost reduction, model routing, provider configuration, context management, cron automation, sub-agent architecture, skills, or agent personality/identity optimization (SOUL.md, IDENTITY.md, AGENTS.md, USER.md audit). Also use when troubleshooting any OpenClaw error, startup failure, channel issue, or provider problem. CLI-first. Advisory by default — audit first, propose exact changes, apply only on approval. Output: prioritized plan + exact CLI commands + config patches + rollback.

- Skill: `modbender/openclaw-optimizer` (Agent Skill, multi-file: 12 files)
- Install (CLI): `npx skillmds@latest add modbender/openclaw-optimizer`
- Raw SKILL.md: https://api.skillmd.com/api/skills/modbender/openclaw-optimizer/raw
- Safety review: pending (external: skill-scanner PASS, skillspector FAIL)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: modbender (https://skillmd.com/u/modbender)
- Updated: 2026-09-09
- Page: https://skillmd.com/skills/modbender/openclaw-optimizer

---


# OpenClaw Optimizer

**Aligned with: OpenClaw v2026.2.26** | Skill v1.16.0 | Updated: 2026-02-27 | CLI-first advisor

Optimize and troubleshoot OpenClaw workspaces: cost-aware routing, provider configuration, context discipline, lean automation, multi-agent architectures, and error resolution.

**Reference files (load when needed):**
- `references/providers.md` — all 29 providers, custom provider schema, failover config
- `references/troubleshooting.md` — full error reference, 7 failure categories, GitHub issue workarounds
- `references/cli-reference.md` — complete CLI command reference
- `references/identity-optimizer.md` — agent identity/personality audit checklist, file roles, walkthrough workflow

---

## Version Awareness

This skill tracks OpenClaw releases via two mechanisms:

1. **GitHub Actions** — daily workflow checks for new releases, opens an issue on drift, auto-closes when resolved
2. **Runtime check** — lightweight cached version comparison at session start

### Runtime Check (once per session)

```bash
python3 ~/.claude/skills/openclaw-optimizer/scripts/version-check.py --status
```

- **`CURRENT`** → note the version and proceed.
- **`STALE`** → inform the user: "OpenClaw v`<new>` is available (skill is at v`<current>`). Run `update-skill.sh` to review what changed."
- **`UNCHECKED`** → note "Version check unavailable (offline)" and proceed.

### Update Workflow (user-initiated, never automatic)

```bash
# Show drift report, changelog, and affected sections
bash ~/.claude/skills/openclaw-optimizer/scripts/update-skill.sh

# After updating content in SKILL.md and references/:
bash ~/.claude/skills/openclaw-optimizer/scripts/update-skill.sh --apply    # bump versions
bash ~/.claude/skills/openclaw-optimizer/scripts/update-skill.sh --commit   # bump + commit + push
```

Updates are deliberate — this skill never auto-modifies its own content or pushes to git without explicit user action.

---

## Quick Start (copy/paste prompts)

**Full audit (safe, no changes):**
> Audit my OpenClaw setup for cost, reliability, and context bloat. Prioritized plan with rollback. Do NOT apply changes.

**Troubleshoot a specific problem:**
> [Describe your symptom or paste the error message]. Diagnose it and give me the exact fix.

**Add or configure a provider:**
> Add [provider name] as a model provider. Walk me through the CLI steps and show me the exact config before applying.

**Model routing optimization:**
> Propose a tiered routing plan: cheap for heartbeats/cron, mid for daily tasks, premium for coding/reasoning. Exact config + rollback. Do NOT apply.

**Silent cron job:**
> Create a cron job that runs [task] every [interval]. Isolated session, NO_REPLY on nothing-to-do. Show me the command first.

**Audit agent personality & identity:**
> Audit my agent's personality and identity files. Check for conflicts, bloat, and bad practices. Walk me through improvements.

---

## Safety Contract (non-negotiable)

- This skill is **advisory by default** — not an autonomous control-plane.
- **Never** mutate config (`config.apply`, `config.patch`), cron jobs, or persistent settings without explicit user approval.
- Before any approved change: show (1) exact CLI command or config patch, (2) expected impact, (3) rollback command.
- If an optimization reduces monitoring coverage, present Options A/B/C and require the user to choose.

---

## Backup Strategy

Three backup layers exist — don't stack manual backups on top unnecessarily:

| Layer | What | Retention | When It's Enough |
|---|---|---|---|
| **CLI rolling `.bak`** | Auto-created on every `config set`, `models set`, `cron edit` | Rolling (overwritten each write) | Single-command undo |
| **Nightly GitHub backup** | Full config committed by cron job (3 AM) | Git history (unlimited) | Any rollback to a previous day's state |
| **Manual dated backup** | `cp <file> <file>.YYYY-MM-DD-<reason>` | Until next nightly covers it, then delete | Major upgrades, multi-file restructuring, direct JSON edits |

**Rule:** For routine CLI changes (model swaps, cron edits, config sets), do NOT create manual backups. The CLI `.bak` + nightly GitHub backup are sufficient. Only create a manual backup when: (1) upgrading OpenClaw versions, (2) editing multiple config files simultaneously (identity audits), or (3) editing JSON directly without the CLI.

---

## 1. Model Providers

29 providers supported. For full docs (auth commands, config schemas, all model names, custom provider setup): **read `references/providers.md`**

**Quick lookup — slug, auth env, primary model format:**

| Provider | Slug | Auth Env | Model Format |
|---|---|---|---|
| Anthropic | `anthropic` | `ANTHROPIC_API_KEY` | `anthropic/claude-opus-4-6` |
| OpenAI (API key) | `openai` | `OPENAI_API_KEY` | `openai/gpt-5.1-codex` |
| **OpenAI Codex (subscription)** | `openai-codex` | **ChatGPT OAuth** | `openai-codex/gpt-5.3-codex` |
| Google Gemini | `google` | `GEMINI_API_KEY` | `google/gemini-3-pro-preview` |

> **WARNING (Feb 2026):** Google is actively cracking down on Gemini CLI OAuth (Antigravity) access through third-party tools. Accounts are being banned or rate-limited. Use API key auth (`google` provider) instead of OAuth (`google-gemini-cli`). Production API keys: 150-300 RPM, no ban risk. See GitHub Issue #14203.

| Mistral | `mistral` | `MISTRAL_API_KEY` | `mistral/mistral-large-latest` |
| Groq | `groq` | `GROQ_API_KEY` | `groq/<model-id>` |
| xAI | `xai` | `XAI_API_KEY` | `xai/grok-code-fast-1` |
| OpenRouter | `openrouter` | `OPENROUTER_API_KEY` | `openrouter/anthropic/claude-sonnet-4-5` |
| Bedrock | `amazon-bedrock` | AWS env chain | `amazon-bedrock/us.anthropic.claude-opus-4-6-v1:0` |
| **Kilo Gateway** | `kilocode` | `KILOCODE_API_KEY` | `kilocode/anthropic/claude-opus-4.6` |
| Moonshot/Kimi | `moonshot` | `MOONSHOT_API_KEY` | `moonshot/kimi-k2.5` |
| Z.AI / GLM | `zai` | `ZAI_API_KEY` | `zai/glm-5` |
| MiniMax | `minimax` | `MINIMAX_API_KEY` | `minimax/MiniMax-M2.1` |
| Venice AI | `venice` | `VENICE_API_KEY` | `venice/llama-3.3-70b` |
| Synthetic | `synthetic` | `SYNTHETIC_API_KEY` | `synthetic/hf:MiniMaxAI/MiniMax-M2.1` |
| Ollama (local) | `ollama` | `OLLAMA_API_KEY` (any) | `ollama/llama3.3` |
| vLLM (local) | `vllm` | `VLLM_API_KEY` (any) | `vllm/<model-id>` |

**Add a provider (API key):**
```bash
openclaw onboard --auth-choice <provider>-api-key
openclaw models auth login --provider <slug>
openclaw models set <provider/model>
```

**Add a provider (OAuth / subscription):**
```bash
openclaw onboard --auth-choice openai-codex    # ChatGPT subscription
openclaw models auth login --provider openai-codex
openclaw models set openai-codex/gpt-5.3-codex
```

### OAuth Providers (Subscription-Based Access)

Some providers offer OAuth authentication tied to a consumer subscription (e.g., ChatGPT Plus/Pro) instead of — or in addition to — a pay-per-token API key. OpenClaw supports these via device-flow OAuth.

**Currently supported OAuth providers:**

| Provider | Slug | Subscription Required | Top Models |
|---|---|---|---|
| **OpenAI Codex** | `openai-codex` | ChatGPT Plus ($20/mo) or Pro ($200/mo) | `gpt-5.3-codex`, `gpt-5.2-codex`, `codex-mini-latest` |
| GitHub Copilot | `github-copilot` | Copilot subscription | `github-copilot/gpt-4o` |

**OpenAI Codex setup (full walkthrough):**

```bash
# 1. Authenticate (opens browser for ChatGPT sign-in)
openclaw models auth login --provider openai-codex
# → Prints a URL. Open it in a browser, sign in to ChatGPT, paste redirect URL back.

# 3. Verify auth
openclaw models status --probe --probe-provider openai-codex

# 4. Set as primary OR add to fallback chain
openclaw models set openai-codex/gpt-5.3-codex          # as primary
openclaw models fallbacks add openai-codex/gpt-5.3-codex # or as fallback

# 5. Restart gateway
launchctl kickstart -k gui/$(id -u)/ai.openclaw.gateway  # macOS LaunchAgent
```

**Headless / SSH gateway:** The OAuth flow prints a URL. Open it in any browser (doesn't need to be on the gateway machine), complete sign-in, then paste the redirect URL back into the SSH terminal. Alternatively, complete OAuth on a machine with a browser and copy `~/.openclaw/credentials/oauth.json` to the gateway.

**Available Codex models:**

| Model | Plan | Notes |
|---|---|---|
| `openai-codex/gpt-5.3-codex` | Plus, Pro, Business | Latest (Feb 2026), most capable coding model |
| `openai-codex/gpt-5.3-codex-spark` | **Pro only** | Research preview, low-latency |
| `openai-codex/gpt-5.2-codex` | Plus, Pro | Previous gen, stable |
| `openai-codex/codex-mini-latest` | Plus, Pro | Lightweight, fast, cheapest |

**Usage limits (per 5-hour window):**
- **Plus ($20/mo):** 30–150 messages
- **Pro ($200/mo):** 300–1,500 messages
- Extra credits purchasable when limits are hit

**Gotchas:**
- **No embeddings.** Codex OAuth does NOT grant access to OpenAI embeddings. You still need a separate `OPENAI_API_KEY` for `text-embedding-3-small` etc.
- **Token refresh is automatic** — active sessions continue without re-login. Credentials stored in `~/.openclaw/credentials/oauth.json`.
- **Don't use both Codex CLI and OpenClaw simultaneously** — some providers invalidate older refresh tokens when a new one is issued. Logging in via one tool can log you out of the other.
- **"Model not supported" errors** — some users report this with `gpt-5.3-codex` on certain accounts. Fall back to `gpt-5.2-codex` if this happens.
- **Dual-config registration required (Issue #13189):** The built-in catalog uses wrong API type (`openai-completions`) for gpt-5.3-codex. Must register manually in **both** `models.json` (API type: `openai-codex-responses`) **AND** `openclaw.json` (API type: `openai-responses` — `openai-codex-responses` is only valid in models.json per schema). v2026.2.26 includes a schema fix — verify with `openclaw models status --probe` after upgrade.
- **Community context (Feb 2026):** After Anthropic and Google updated their ToS to block subscription-based OAuth in third-party tools, the OpenClaw community migrated heavily to `openai-codex`. OpenAI explicitly permits Codex OAuth in third-party tools, though fair-use limits still apply.

### Provider Removal Checklist

Removing a provider requires cleaning **6 locations** — `config unset` alone is not enough:

1. `models.providers.<slug>` in openclaw.json — `openclaw config unset models.providers.<slug>`
2. `auth.profiles.<slug>:*` in openclaw.json — must edit JSON directly (colons in keys break `config unset`)
3. `profiles` dict in `~/.openclaw/agents/main/agent/auth-profiles.json` — edit with python3/jq
4. `agents.defaults.models.<provider/model>` aliases in openclaw.json — `openclaw config unset` each alias
5. `plugins.entries.<slug>-auth` in openclaw.json — `openclaw config unset plugins.entries.<slug>-auth`
6. `lastGood.<slug>` and `usageStats.<slug>:*` in auth-profiles.json — edit directly

For providers with LaunchAgent env vars (Ollama, etc.), also clean:
7. `launchctl unsetenv <KEY>` — session-level env persists independently of plist
8. PlistBuddy delete from `~/Library/LaunchAgents/ai.openclaw.gateway.plist`
9. `launchctl bootout` + `launchctl bootstrap` to pick up the clean plist (kickstart alone doesn't reload env from plist)

**Known CLI limitation:** `openclaw config unset` cannot handle colons in config keys (e.g., `auth.profiles.google-gemini-cli:email@gmail.com`). The parser treats colons as path separators. Edit the JSON file directly for these entries.

**Custom OpenAI-compatible provider (LM Studio, LiteLLM, etc.):** See `references/providers.md`

---

## 2. Model Routing Strategy

### Tiered Routing (50–95% cost reduction)

| Tier | Models | Use Cases |
|---|---|---|
| **T1 Cheap** | `zai/glm-5`, `google/gemini-3-flash-preview`, `synthetic/hf:deepseek-ai/DeepSeek-V3.2` | Heartbeats, simple checks, greetings, cron |
| **T2 Mid** | `moonshot/kimi-k2.5`, `minimax/MiniMax-M2.1` | Daily chat, Q&A, calendar, scheduling |
| **T3 Smart** | `anthropic/claude-sonnet-4-5`, `openai/gpt-5.1-codex`, `openai-codex/gpt-5.3-codex` (subscription) | Code, refactors, research |
| **T4 Premium** | `anthropic/claude-opus-4-6`, `openai/gpt-5.2` | Complex reasoning, orchestration |

**Model preference by task:**

| Task | Model | Why |
|---|---|---|
| Heartbeats / cron | `zai/glm-5` | Cheapest; reliable structured output |
| Calendar / scheduling | `moonshot/kimi-k2.5` | Community #1 for date/time reasoning |
| Coding / refactoring | `anthropic/claude-sonnet-4-5` or `openai-codex/gpt-5.3-codex` | Sonnet: community #1 for code quality; Codex: flat-rate via subscription |
| Agent orchestration | `anthropic/claude-opus-4-6` | Best multi-step reasoning |
| Long-context tasks | `google/gemini-3-flash-preview` | 1M token window |
| Subscription-capped coding | `openai-codex/gpt-5.3-codex` | Fixed cost via ChatGPT Plus/Pro; no per-token billing |
| Privacy-sensitive | `venice/llama-3.3-70b` or Ollama | Never logged/stored |

**Key rules:**
- Never switch models mid-conversation — destroys Anthropic prompt cache
- Use `anthropic` direct (not through proxies) to preserve caching for Opus/Sonnet
- Switch only at session boundaries (`/new`)

### Per-Agent Config

```bash
openclaw models set anthropic/claude-opus-4-6           # set global primary
openclaw config set agents.defaults.model.primary anthropic/claude-opus-4-6
openclaw models fallbacks add openrouter/anthropic/claude-sonnet-4-5
```

```json5
{
  agents: {
    list: [
      { id: "main", model: "anthropic/claude-opus-4-6", heartbeat: { every: "30m" } },
      { id: "ops",  model: { primary: "anthropic/claude-sonnet-4-5", fallbacks: ["zai/glm-5"] },
        tools: { profile: "minimal" } },
    ],
    defaults: {
      model: { primary: "anthropic/claude-opus-4-6", fallbacks: ["minimax/MiniMax-M2.1"] },
      thinkingDefault: "low",
      timeoutSeconds: 600,
      contextTokens: 200000,
      maxConcurrent: 3,
      params: { cacheRetention: "long" },   // per-agent param overrides (v2026.2.23+)
    },
  },
}
```

**In-chat model switch (no restart):** `/model list` → `/model anthropic/claude-sonnet-4-5`

---

## 3. Context Management

**What burns tokens:** System prompt (5–10K tokens/call) + bootstrap files + conversation history. Bootstrap files injected on every turn (source: `docs.openclaw.ai/concepts/system-prompt`): `AGENTS.md`, `SOUL.md`, `TOOLS.md`, `IDENTITY.md`, `USER.md`, `HEARTBEAT.md`, `BOOTSTRAP.md` (first-run only), plus `MEMORY.md` and/or `memory.md` **when present**. Daily `memory/*.md` files are NOT auto-injected (on-demand via memory tools). Bootstrap cap: 150K chars total, 20K per file (both configurable).

> **MEMORY.md warning (from docs):** *"Keep them concise — especially MEMORY.md, which can grow over time and lead to unexpectedly high context usage and more frequent compaction."* MEMORY.md is the most common source of bootstrap bloat. Unlike AGENTS.md or SOUL.md which users actively edit, MEMORY.md tends to grow unchecked as the agent appends to it.

**Check context:** `/status` · `/context list` · `/context detail` · `/usage tokens`

### Compaction Config

```bash
# Manual: /compact [focus instructions]
# Auto: triggers near context limit — count visible in /status

openclaw config set agents.defaults.compaction.mode safeguard
openclaw config set agents.defaults.compaction.reserveTokensFloor 32000
openclaw config set agents.defaults.contextTokens 100000
```

```json5
{
  agents: { defaults: { compaction: {
    mode: "safeguard",
    reserveTokensFloor: 32000,
    memoryFlush: {
      enabled: true,
      prompt: "Write lasting notes to memory/YYYY-MM-DD.md; reply NO_REPLY if nothing to store.",
    },
  }, contextTokens: 100000 } },
}
```

**Known bug — memory flush threshold gap (Issue #25880):** Set `reserveTokensFloor` equal to `reserveTokens` (both `62500`) to fix compaction firing before flush completes.

### Bootstrap File Size Targets (optimization recommendations)

These are optimization targets for keeping context lean, not hard limits. All files are subject to `bootstrapMaxChars` (default 20K) and `bootstrapTotalMaxChars` (default 150K).

| File | Target Size | Purpose | Injected? |
|---|---|---|---|
| `SOUL.md` | < 1K tokens (~4K chars) | Personality + absolute constraints | Always (main + full prompt mode) |
| `AGENTS.md` | < 2K tokens (~8K chars) | Workflows, rules, operating procedures | Always (main + sub-agents) |
| `TOOLS.md` | < 2K tokens (~8K chars) | Tool-specific notes, local conventions | Always (main + sub-agents) |
| `IDENTITY.md` | < 500 tokens (~2K chars) | Name, vibe, emoji, presentation | Always (main only) |
| `USER.md` | < 1K tokens (~4K chars) | User profile, preferences, context | Always (main only) |
| `HEARTBEAT.md` | < 200 tokens (~800 chars) | Heartbeat checklist (keep minimal) | Always (main only) |
| `MEMORY.md` | < 5K tokens (~20K chars) | **Curated long-term facts ONLY** | **Always in main sessions (auto-injected when present)** |

**Critical:** MEMORY.md is auto-injected on every turn in main sessions, NOT loaded on-demand. It burns tokens continuously. Keep it as small as possible with only curated facts. Operational protocols belong in AGENTS.md. Tool notes belong in TOOLS.md.

### Bootstrap Content Placement (What Goes Where)

Users commonly dump all content into SOUL.md because it feels like "who the agent is." This bloats the file (burns tokens every turn) and confuses lighter models that can't prioritize across a noisy instruction set. Place content in the correct file:

| Content Type | Correct File | Common Mistake |
|---|---|---|
| Personality, voice, humor, constraints | SOUL.md | - |
| Protocols, workflows, checklists, operational rules | AGENTS.md | Dumping in SOUL.md |
| User bio, preferences, working hours, communication style | USER.md | Duplicating in SOUL.md |
| Tool configs, API templates, channel IDs, env vars | TOOLS.md | Scattering in AGENTS.md |
| Curated long-term facts (lean) | MEMORY.md | Growing unchecked |
| Proactivity rules, initiative behavior | AGENTS.md | Putting in SOUL.md |

**Cross-file duplication burns tokens silently.** If the same protocol appears in both SOUL.md and AGENTS.md, it's injected twice on every turn. Deduplicate aggressively — pick one canonical location.

**Stale model references are silent saboteurs.** When you change models via CLI (`openclaw models set`), update any AGENTS.md sections that reference specific model names (e.g., Model Selection, Sub-Agent defaults). The agent follows bootstrap instructions and may try to use models that are no longer configured.

**Persistence stack:** `SOUL.md` → `AGENTS.md` → `TOOLS.md` → `IDENTITY.md` → `USER.md` → `MEMORY.md` (all auto-injected in main sessions) → `memory/YYYY-MM-DD.md` (on-demand via memory tools) → `conversation-state.md` → `ACTIVE-TASK.md`

### Session Maintenance

```bash
openclaw config set session.maintenance.mode enforce
openclaw config set session.maintenance.maxDiskBytes 500mb
openclaw sessions cleanup --dry-run      # preview
openclaw sessions cleanup --enforce     # apply
openclaw sessions cleanup --fix-missing # prune store entries whose transcript files are missing (v2026.2.26+)
```

---

## 4. Cron & Automation

### Cron Job Schema (key fields)

```json
{
  "jobId": "daily-brief", "name": "Morning Briefing", "enabled": true,
  "agentId": "main",
  "schedule": { "kind": "cron", "expr": "0 8 * * *", "tz": "America/New_York" },
  "sessionTarget": "isolated",
  "payload": { "kind": "agentTurn", "message": "Morning briefing.", "model": "anthropic/claude-sonnet-4-5", "timeoutSeconds": 300 },
  "delivery": { "mode": "announce", "channel": "telegram", "to": "<user-id>" }
}
```

**sessionTarget:** `"isolated"` (recommended — fresh session) | `"main"` (injects as systemEvent)
**payload.kind:** `"agentTurn"` (isolated) | `"systemEvent"` (main session)
**delivery.mode:** `"announce"` | `"webhook"` | `"none"`

### CLI

```bash
openclaw cron add --cron "0 9 * * *" --message "Daily report" --agent main --announce --channel slack --to "channel:CXXX"
openclaw cron add --at "2026-03-01T08:00:00" --message "One-time task" --keep-after-run
openclaw cron run <job-id>          # test immediately (--force bypasses not-due)
openclaw cron list / status / runs
openclaw cron edit <job-id> [flags] # patch fields: --cron, --message, --model, --name, --tz, etc.
openclaw cron enable/disable <job-id>
openclaw cron rm <job-id>
openclaw config set cron.sessionRetention 24h
openclaw config set cron.maxConcurrentRuns 1   # circuit breaker
```

### Silent Patterns

**`NO_REPLY`** — agent outputs this literal string when nothing to report; system suppresses delivery entirely.
**`HEARTBEAT_OK`** — heartbeat token; reply ≤300 chars after stripping it → silently dropped.

```json5
{
  agents: { defaults: { heartbeat: {
    every: "30m",
    target: "last",
    ackMaxChars: 300,
    directPolicy: "allow",   // "allow" (default in v2026.2.25+) | "block" — per-agent override also supported
    activeHours: { start: "08:00", end: "22:00", timezone: "America/New_York" },
  } } },
}
```

> **v2026.2.25 BREAKING:** The heartbeat DM toggle was replaced with `directPolicy`. Default is now `allow`. If you had DMs blocked in v2026.2.24, explicitly set `agents.defaults.heartbeat.directPolicy: "block"` (or per-agent via `agents.list[].heartbeat.directPolicy`).

**Cost trap:** 5-minute heartbeat loading full MEMORY.md = ~2.9M tokens/day. Keep heartbeat context minimal.

**Redundant cron jobs:** The built-in `openclaw memory` indexes sessions natively. Custom session archiver cron jobs that convert `.jsonl` to markdown for a separate RAG database are likely redundant. Check whether any cron job feeds a custom system that duplicates built-in functionality before assuming it's needed.

**Known bugs:** Cron current-day skip (Issue #25902) — restart the gateway with `launchctl kickstart -k gui/$(id -u)/ai.openclaw.gateway` to recompute (do NOT use `openclaw gateway restart` — it causes duplicate processes; see Section 10). Cron announce → Telegram failure (Issue #25906) — switch to `directMessage` mode.

**v2026.2.25 fixes:** Cron model override failures now auto-recover — if an isolated job's `payload.model` is no longer allowlisted, it gracefully falls back to the default model instead of failing the job. Cron announce duplicate sends are also fixed (duplicate guard tracks attempted vs confirmed delivery). Multi-account cron routing now properly honors `delivery.accountId`.

---

## 5. Skills & Plugins

**`metadata.openclaw.requires`** — gates skill visibility:
```yaml
metadata:
  openclaw:
    requires:
      bins: ["ffmpeg"]          # ALL must exist on PATH
      anyBins: ["gh", "hub"]    # AT LEAST ONE must exist
      env: ["GITHUB_TOKEN"]     # env vars that must be set
      config: ["browser.enabled"]
    os: ["darwin", "linux"]
```

**`disable-model-invocation: true`** — removes skill from model's tool list; user can still invoke manually. Use for high-impact or security-sensitive skills.

**Skills directory:** `~/.openclaw/workspace/skills/` — this is the filesystem path where all skills are stored. Each skill lives in its own subdirectory (e.g., `~/.openclaw/workspace/skills/my-skill/SKILL.md`). When manually installing or copying skills, always use this path — not `~/.openclaw/skills/`.

**ClawHub:**
```bash
npx clawhub install <slug>       # install
clawhub update --all             # update all
openclaw skills list --eligible  # what's loaded
openclaw skills check            # validate requirements
```

**Security:** Before installing any skill, read its `SKILL.md` manually. Community scans found 341 malicious skills (reverse shells, credential exfiltration). New accounts with popular skills = red flag.

**Session watcher:** Skills snapshot at session start. If `skills.load.watch` is disabled, start a new session after installing.

---

## 6. Multi-Agent & Sub-Agent Architecture

```bash
/subagents spawn ops "Audit logs from last 24h"   # via chat
```

```json5
// sessions_spawn tool (programmatic)
{ "task": "Audit logs", "agentId": "ops", "model": "anthropic/claude-sonnet-4-5",
  "thinking": "low", "runTimeoutSeconds": 300, "mode": "minimal" }
```

```json5
// Nesting config
{ agents: { defaults: { subagents: {
  maxSpawnDepth: 2,    // 0=off; 1=spawn; 2=orchestrator
  maxConcurrency: 8,
  maxChildrenPerAgent: 5,
} } } }
```

**Community pattern:** Orchestrator (`opus-4-6`) → Code sub-agent (`sonnet-4-5`) → Research sub-agent (`kimi-k2.5`) → Cron/monitoring (`zai/glm-5`, isolated)

**Sandbox isolation:**
```json5
{ agents: { list: [{ id: "untrusted", sandbox: { mode: "docker" },
  tools: { profile: "minimal", deny: ["exec", "browser"] } }] } }
```

---

## 7. High-ROI Optimization Levers

| Lever | Impact | How |
|---|---|---|
| **Tiered model routing** | 50–95% cost reduction | T1 for cron/heartbeat, T4 only for orchestration |
| **Prompt caching** | 60–90% input token reduction | Keep system prompt stable; use `anthropic` direct |
| **Bootstrap file discipline** | 2K–10K tokens/call saved | SOUL.md <1K, AGENTS.md <2K, MEMORY.md <5K |
| **Silent cron (NO_REPLY)** | Eliminates delivery tokens | Instruct: "Reply NO_REPLY if nothing actionable" |
| **Compaction tuning** | Prevents overflow disasters | `safeguard` mode, `reserveTokensFloor: 32000` |
| **Session maintenance** | Prevents disk/perf degradation | `mode: enforce`, `maxDiskBytes: 500mb` |
| **Batch heartbeat checks** | 10x fewer API calls | One heartbeat for 10 checks > 10 cron jobs |
| **Isolated cron sessions** | Zero context contamination | `sessionTarget: "isolated"` on all cron jobs |
| **Gateway security** | Prevents exposure | `gateway.bind: loopback`; Tailscale for remote |
| **Never switch mid-session** | Preserves prompt cache | Only switch model at `/new` boundaries |

---

## 8. CLI Reference

**Best practice (v2026.2.25+):** Before editing config or asking config-field questions, have the agent call the `config.schema` tool in-chat. This returns the current schema with valid keys, types, and defaults — avoids guessing or using stale field names. Note: this is an agent in-chat tool, NOT a CLI command.

**Most common commands:**
```bash
openclaw doctor --fix               # auto-fix config issues
openclaw gateway status             # check runtime + RPC probe
openclaw models set <provider/model>
openclaw models status --probe
openclaw cron run <job-id>          # test a cron job immediately
openclaw sessions cleanup --dry-run
openclaw sessions cleanup --fix-missing  # prune entries with missing transcripts (v2026.2.26+)
openclaw update
openclaw security audit             # post-upgrade check
openclaw secrets audit              # scan bootstrap files for hardcoded secrets (v2026.2.26+)
openclaw secrets configure          # configure external secrets (v2026.2.26+)
openclaw secrets apply              # apply secrets with strict target-path validation (v2026.2.26+)
openclaw agents bindings            # list account-scoped agent route bindings (v2026.2.26+)
openclaw agents bind                # bind agent to channel account (v2026.2.26+)
openclaw agents unbind              # unbind agent from channel account (v2026.2.26+)
```

> **`openclaw onboard --reset` scope change (v2026.2.26):** Default reset scope is now `config+creds+sessions`. Workspace deletion (bootstrap files, skills, memory) now requires `--reset-scope full`. Do NOT run `openclaw onboard --reset` without specifying `--reset-scope` explicitly — the default no longer wipes the workspace.

**Gateway restart (macOS LaunchAgent):**
```bash
# SAFE restart — single atomic operation, no duplicate processes
launchctl kickstart -k gui/$(id -u)/ai.openclaw.gateway

# DO NOT use `openclaw gateway restart` — it races with KeepAlive and spawns
# duplicate processes that loop "Port already in use" every ~10s at 100%+ CPU.

# Recovery if duplicates already exist:
launchctl bootout gui/$(id -u)/ai.openclaw.gateway    # stop launchd service + kill managed process
kill <any-remaining-pids>                              # kill orphans
launchctl bootstrap gui/$(id -u) ~/Library/LaunchAgents/ai.openclaw.gateway.plist  # re-register + start clean
```

**Full CLI reference (all commands, flags, in-chat commands):** Read `references/cli-reference.md`

---

## 9. Ops Hygiene Checklist

**Daily:**
- `openclaw health --json` via cron (→ HEARTBEAT_OK if clean)
- `clawhub whoami` to verify ClawHub auth
- Token budget check (cost-sensitive providers)

**Weekly:**
- `openclaw update --dry-run` → review → `openclaw update`
- `clawhub update --all --dry-run` → review → `clawhub update --all`
- Curate MEMORY.md — archive old daily logs, promote key insights
- `openclaw sessions cleanup --dry-run` → `openclaw sessions cleanup`
- `openclaw cron status` — check for errors
- Clean stale backup files: `find ~/.openclaw -name "*.bak.*" -mtime +7 -not -name "*.bak" | xargs rm -v` (preserves CLI's rolling `.bak` files, removes old named/dated backups)

**Quarterly:**
- Review custom scripts (`scripts/`) for redundancy with built-in OpenClaw features. Users often build custom solutions (RAG pipelines, session archivers, memory indexers) that become redundant when OpenClaw adds equivalent built-in functionality. Check whether each script and its associated cron job still serves a purpose that the platform doesn't already handle.

**Before/After Updates:**
- After update: `openclaw doctor --fix` (handles config migrations automatically)
- v2026.2.23 breaking change: `allowPrivateNetwork` → `dangerouslyAllowPrivateNetwork` — auto-fixed by doctor
- Manual backup only needed for major upgrades or multi-file restructuring (see Backup Strategy below)

**On Every System Assessment (mandatory data collection):**
- `openclaw cron list` + read `~/.openclaw/cron/jobs.json` — capture full cron inventory: job IDs, names, schedules, **model overrides** (from `payload.model`), status, last run times
- Flag any jobs in `error` state — these are active problems
- Flag jobs with stale last-run times (>24h for daily jobs) — may indicate silent failures
- Check timezone consistency — jobs using `(exact)` instead of named timezones may fire at wrong times
- Record whether jobs use `isolated` or `main` session target
- Map cron schedule to day/night distribution — heavy jobs should cluster overnight
- Document all findings in the system profile's `## Cron Jobs` section before making recommendations
- Without this data, recommendations will duplicate existing automation and waste time

**Security:**
- `openclaw config get gateway.bind` → must be `loopback`
- No public port exposure — use Tailscale for remote
- API keys not in skill files or version control
- Audit ClawHub skills before installing

---

## 10. Troubleshooting

**Log file paths (macOS):**
- **Error log:** `~/.openclaw/logs/gateway.err.log` — primary source for errors, 502s, plugin failures, tool errors
- **Main log:** `/tmp/openclaw/openclaw-YYYY-MM-DD.log` — verbose debug output (lane events, session activity)

Always check `gateway.err.log` first when troubleshooting — it contains only errors and warnings, making root cause identification much faster than grepping the main log.

**First — always run this triage sequence:**
```bash
openclaw status
openclaw gateway status            # must show "Runtime: running" + "RPC probe: ok"
openclaw doctor
openclaw channels status --probe
tail -50 ~/.openclaw/logs/gateway.err.log | grep -v DEP0040   # skip Node deprecation noise
```

**Quick fix by symptom:**

| Symptom | First Command | Most Likely Fix |
|---|---|---|
| No response from agent | `openclaw gateway status` | Gateway not running or pairing pending |
| Gateway won't start | `openclaw logs --follow` | `EADDRINUSE` or `gateway.mode` not set to `local` |
| "Port already in use" loop | `ps aux \| grep openclaw-gateway` | Duplicate processes from CLI restart vs LaunchAgent `KeepAlive`. Fix: `launchctl bootout` → kill orphans → `launchctl bootstrap` (see Section 8) |
| "unauthorized" on Control UI | `launchctl getenv OPENCLAW_GATEWAY_TOKEN` | Remove stale launchctl env override |
| Cron job never fires | `openclaw cron status` | Cron disabled or timezone mismatch |
| Heartbeat always skipped | `openclaw config get agents.defaults.heartbeat.activeHours` | Wrong timezone, outside active hours, or `directPolicy` set to `block` (v2026.2.25 changed default to `allow`) |
| Cron job fails with "model not allowlisted" | `openclaw cron status` | v2026.2.25+ auto-recovers by falling back to default model. On older versions: update `payload.model` in the job or re-add the model to the allowlist. |
| Channel message dropped | `openclaw logs --follow` | Mention required or sender not paired |
| "RPC probe: failed" | `openclaw gateway status --deep` | Auth token mismatch or port conflict |
| Post-upgrade breakage | `openclaw doctor --fix` | Automatic config migration |
| Provider 401 errors | `openclaw models status --probe` | Token expired or wrong key type |
| Chrome browser won't start (Linux) | `openclaw browser status` | Snap Chromium conflict → install Google Chrome .deb |
| ALL providers timeout simultaneously | `grep "delivery-recovery" gateway.err.log` | **Not a provider issue.** Two common causes: (A) **Context bloat** — `contextTokens` unset (unlimited), payload too large for any provider to process within `timeoutSeconds`. Fix: set `contextTokens: 100000`, `timeoutSeconds: 180`, `reserveTokensFloor: 32000`. See Section 10d. (B) **Event loop overload** — stuck delivery-queue, skills-remote probes, Gemini OAuth cycling, too many concurrent sessions. Fix: clear delivery queue, set `cron.maxConcurrentRuns: 1`. See Section 10b. |
| Delivery recovery loop ("21 entries deferred") | `ls ~/.openclaw/delivery-queue/` | Stuck entries (wrong channel, message too long) retry forever on every restart. Move to `~/.openclaw/delivery-queue/failed/` to stop the loop. |
| Ollama "fetch failed" (instant, ~100ms) | Check gateway err log for `Failed to discover Ollama models` | **Known bug:** OpenClaw hardcodes `127.0.0.1:11434` for Ollama discovery (Issue #8663). On macOS, LaunchAgent processes are sandboxed and can't reach private LAN IPs like `192.168.x.x` (Issue #21494). Fix: reverse SSH tunnel from Ollama machine to gateway (`ssh -fN -R 127.0.0.1:11434:127.0.0.1:11434 user@gateway`), set `baseUrl` to `http://127.0.0.1:11434`, add `OLLAMA_HOST` and `OLLAMA_API_KEY` to LaunchAgent env. See Section 10a below. |
| Ollama "Connection error" | Same as above | Same root cause. Switching `api` from `ollama` to `openai-completions` changes the error message but doesn't fix it — the sandbox blocks all LAN connections. |
| Ollama probes spike memory | `curl http://host:11434/api/ps` | Set `OLLAMA_KEEP_ALIVE=0` on the Ollama machine so models unload immediately after probes. No OpenClaw config to disable probes per-provider. |
| Gemini CLI "API rate limit reached" | `openclaw logs \| grep rate` | Google OAuth crackdown (Feb 2026). Switch to API key auth. See Section 1 warning. |
| Provider removal didn't stop probes | Check all 6 locations in Provider Removal Checklist | Stale auth-profiles.json, launchctl env, or plist env vars. See Section 1. |
| `config unset` fails on auth profile keys | Edit JSON directly | Colons in keys break the config path parser. Use python3/jq. |
| `models status --probe` mass timeouts | Test individual providers with `curl` | Probe contention — 16+ simultaneous targets saturate the event loop. Not real failures. |

### 10a. Remote Ollama on macOS (Known Bug Workaround)

**Problem:** OpenClaw on macOS cannot connect to a remote Ollama server on the LAN. `curl` works, but the gateway process fails with "fetch failed" or "Connection error." This affects all API modes (`ollama` and `openai-completions`).

**Root causes (two bugs stacking):**
1. **Hardcoded localhost discovery** (Issue [#8663](https://github.com/openclaw/openclaw/issues/8663)): OpenClaw always probes `127.0.0.1:11434` for Ollama, ignoring `baseUrl`.
2. **macOS LaunchAgent sandbox** (Issue [#21494](https://github.com/openclaw/openclaw/issues/21494)): The gateway process running under launchd gets `EHOSTUNREACH` for private network IPs (`192.168.x.x`, `10.x.x.x`).

**Fix — reverse SSH tunnel:**
```bash
# Run on the Ollama machine (not the gateway):
ssh -fN -R 127.0.0.1:11434:127.0.0.1:11434 user@gateway-host

# On the gateway:
openclaw config set models.providers.<ollama-slug>.baseUrl 'http://127.0.0.1:11434'
openclaw config set models.providers.<ollama-slug>.api ollama
```

Add to the gateway's LaunchAgent plist:
```xml
<key>OLLAMA_HOST</key>
<string>http://127.0.0.1:11434</string>
<key>OLLAMA_API_KEY</key>
<string>ollama</string>
```

**Important:** Kill any local Ollama on the gateway first — it will conflict with the tunnel on port 11434. Make the tunnel persistent with a LaunchAgent on the Ollama machine.

### 10b. Multi-Provider Timeout Storms (Event Loop Overload)

**Symptom:** ALL providers (Kimi, KiloCode, Google, Anthropic, etc.) timeout simultaneously within a 30-90 minute window, even though they are independent services. `FailoverError: LLM request timed out.` on every model in the fallback chain. May cause gateway crash/restart.

**Root cause:** The gateway's Node.js event loop is saturated by a pile-up of concurrent operations. Outbound HTTPS responses arrive, but the process can't process them before its own timeout timer fires. The providers are NOT down — the gateway can't handle the responses.

**Common overload contributors (check all of these):**

1. **Stuck delivery-recovery queue** — Files in `~/.openclaw/delivery-queue/` that will never succeed (wrong channel, message too long) retry on every restart and periodically. Each retry burns event loop time.
   - **Diagnose:** `ls ~/.openclaw/delivery-queue/*.json | wc -l` and `grep "delivery-recovery" gateway.err.log | tail -20`
   - **Fix:** `mv ~/.openclaw/delivery-queue/*.json ~/.openclaw/delivery-queue/failed/`

2. **Skills-remote bin probes to offline nodes** — Gateway probes paired nodes for skill binary requirements. If nodes don't have the node service running, each probe hangs until timeout.
   - **Diagnose:** `grep "skills-remote.*timed out" gateway.err.log | wc -l`
   - **Fix:** Remove offline nodes from paired devices, or ensure nodes have the node service running.

3. **Google Gemini CLI OAuth account cycling** — If the agent switches to Gemini CLI in-session, it cycles through OAuth accounts. Each expired/slow account hangs for 90 seconds. 6 accounts = up to 540s of hung connections.
   - **Diagnose:** `grep "google-gemini-cli.*timed out" gateway.err.log | tail -10`
   - **Fix:** Ensure OAuth tokens are fresh, or use the `google` (API key) provider instead of `google-gemini-cli` for fallbacks.

4. **No cron concurrency limit** — Multiple cron jobs firing simultaneously all compete for the same event loop and hit the same provider chain, creating a thundering herd.
   - **Fix:** `openclaw config set cron.maxConcurrentRuns 1`

5. **Proxy providers as early fallbacks** — KiloCode is a proxy. When it degrades, ALL models through it fail simultaneously (appears as multiple independent failures but is one SPOF). Put direct-API providers (Anthropic, Google API key) before proxies in the fallback chain.

**Recovery:** After fixi

…(truncated)