Skip to content

SKILL.md

Latest commit

 

History

History
292 lines (196 loc) · 11.8 KB

SKILL.md

File metadata and controls

292 lines (196 loc) · 11.8 KB
Raw
name pp-ollama-cloud
description Routes every prompt to the right hosted Ollama model. Wraps chat, embeddings, catalog, and the OpenAI-compatible... Trigger phrases: `pick an ollama model`, `advise on a model`, `which ollama cloud model should I use`, `route this prompt`, `use ollama cloud`, `run ollama cloud`.
author user
license Apache-2.0
argument-hint <command> [args] | install cli|mcp
allowed-tools Read Bash
metadata
openclaw
requires
bins
ollama-cloud-pp-cli

Ollama Cloud — Printing Press CLI

Prerequisites: Install the CLI

This skill drives the ollama-cloud-pp-cli binary. You must verify the CLI is installed before invoking any command from this skill. If it is missing, install it first:

  1. Install via the Printing Press installer. It defaults binaries to $HOME/.local/bin on macOS/Linux and %LOCALAPPDATA%\Programs\PrintingPress\bin on Windows: 
    npx -y @mvanhorn/printing-press-library install ollama-cloud --cli-only
  2. Verify: ollama-cloud-pp-cli --version
  3. Ensure the reported install directory is on $PATH for the agent/runtime that will invoke this skill.

If the npx install fails (no Node, offline, etc.), fall back to a direct Go install (requires Go 1.26.4 or newer):

go install github.com/mvanhorn/printing-press-library/library/ai/ollama-cloud/cmd/ollama-cloud-pp-cli@latest

If --version reports "command not found" after install, the runtime cannot see the binary directory on $PATH. Do not proceed with skill commands until verification succeeds.

When to Use This CLI

Pick this CLI when you need agent-driven access to Ollama's hosted catalog and want routing intelligence baked in. Especially useful when the alternative is hardcoding a model name and hoping it stays right. Pairs naturally with multi-backend setups where ollama-cloud is one of several providers.

Unique Capabilities

These capabilities aren't available in any other tool for this API.

Routing intelligence

  • advise — Picks the right Ollama Cloud model for a prompt by combining live catalog, heuristic prompt-feature extraction, curated cost/latency metadata, and an optional cheap meta-LLM tiebreak.

    When an agent needs to pick a hosted Ollama model and the default routing is wrong, reach for advise instead of hardcoding the model name.

    ollama-cloud-pp-cli advise --prompt-file ./prompt.txt --task-hint coding --budget-remaining-usd 0.50 --json

  • compare — Runs the same prompt against N hosted models in parallel and emits side-by-side response, tokens, and latency.

    Use when calibrating advisor recommendations or picking between two close models.

    ollama-cloud-pp-cli compare --prompt-file ./p.txt --models qwen3-coder:480b,gpt-oss:120b,deepseek-v3.1:671b --json

  • advise — With --explain, advise emits the full scoring trace: feature extraction, per-model scores, filter passes, tiebreak rationale.

    Reach for this when an advise recommendation surprises you and you want to understand why.

    ollama-cloud-pp-cli advise --prompt-file ./p.txt --explain --format md

Engagement canary

  • advise-replay — Replays advisor recommendations and reports divergence between recommended models and actually-chosen models. Foundation for the divergence canary; the prompt corpus is not retained so judge-LLM scoring is not in scope until a corpus sidecar ships.

    Run weekly to detect advisor drift; surfaces divergence between recommended and actual-chosen models.

    ollama-cloud-pp-cli advise-replay --since 7d --diverge-only --json --select rows,divergence_count,divergence_pct

Operations

  • budget — Probes the free-tier weekly cap with a 1-token chat. Parses Ollama Cloud's 429 prose and emits a structured verdict (ok | exhausted | unknown) with the upgrade URL so agents can pre-flight quota before launching long sessions.

    Run before launching a long agent session to confirm quota is available.

    ollama-cloud-pp-cli budget --json

  • cost-trace — Aggregates advisor-log cost estimates over a time window; compares per-model and per-task-hint spend.

    Use to decide whether to upgrade to a paid Ollama Cloud tier.

    ollama-cloud-pp-cli cost-trace --since 7d --group-by task-hint --json

Command Reference

chat — Manage chat

  • ollama-cloud-pp-cli chat chat — Native Ollama chat endpoint. Supports streaming.
  • ollama-cloud-pp-cli chat completions — OpenAI-compatible chat completions endpoint.

embeddings — Manage embeddings

  • ollama-cloud-pp-cli embeddings embed — Native Ollama embeddings endpoint.
  • ollama-cloud-pp-cli embeddings openai-embed — Generate embeddings (OpenAI-compatible)

models — Manage models

  • ollama-cloud-pp-cli models — Catalog in OpenAI list-models format.

ps — Manage ps

  • ollama-cloud-pp-cli ps — Shows currently-loaded models. On Ollama Cloud this typically reflects models with active sessions.

show — Manage show

  • ollama-cloud-pp-cli show — Returns model metadata, template, modelfile, capabilities.

tags — Manage tags

  • ollama-cloud-pp-cli tags — Returns the live catalog of hosted Ollama Cloud models.

Finding the right command

When you know what you want to do but not which command does it, ask the CLI directly:

ollama-cloud-pp-cli which "<capability in your own words>"

which resolves a natural-language capability query to the best matching command from this CLI's curated feature index. Exit code 0 means at least one match; exit code 2 means no confident match — fall back to --help or use a narrower query.

Recipes

Advise with explain trace as markdown

ollama-cloud-pp-cli advise --prompt-file ./prompt.txt --task-hint coding --explain --format md

--explain populates the feature trace and filtered-candidate audit; --format md renders for humans.

Detect catalog drift

ollama-cloud-pp-cli advise --validate-catalog --json --select uncurated_live,curated_not_in_live

Diffs live /api/tags against curated models.json overlay; flags both uncurated live models and stale curated patterns.

Replay last week recommendations

ollama-cloud-pp-cli advise-replay --since 7d --dry-run --json --select divergence_count,divergence_pct,emitted

Engagement canary; dotted-path select keeps response under context budget.

Side-by-side comparison

ollama-cloud-pp-cli compare --prompt-file ./prompt.txt --models qwen3-coder:480b,gpt-oss:120b --json

Runs the prompt against both models in parallel and reports per-model latency, tokens, and rate-limit state.

Cost rollup by task hint

ollama-cloud-pp-cli cost-trace --since 7d --group-by task-hint --json --select groups,row_count

Aggregates the advisor JSONL log by task-hint so you can decide whether to upgrade the Ollama Cloud tier.

Auth Setup

Bearer auth via OLLAMA_CLOUD_API_KEY (intentionally distinct from any local-Ollama env var). Free tier is rate-limited weekly; the budget command surfaces exhaustion before workflows fail.

Run ollama-cloud-pp-cli doctor to verify setup.

Agent Mode

Add --agent to any command. Expands to: --json --compact --no-input --no-color --yes.

  • Pipeable — JSON on stdout, errors on stderr

  • Filterable — --select keeps a subset of fields. Dotted paths descend into nested structures; arrays traverse element-wise. Critical for keeping context small on verbose APIs:

    ollama-cloud-pp-cli chat chat --model example-value --agent --select id,name,status

  • Previewable — --dry-run shows the request without sending

  • Offline-friendly — sync/search commands can use the local SQLite store when available

  • Non-interactive — never prompts, every input is a flag

  • Explicit retries — use --idempotent only when an already-existing create should count as success

Response envelope

Commands that read from the local store or the API wrap output in a provenance envelope:

{
  "meta": {"source": "live" | "local", "synced_at": "...", "reason": "..."},
  "results": <data>
}

Parse .results for data and .meta.source to know whether it's live or local. A human-readable N results (live) summary is printed to stderr only when stdout is a terminal — piped/agent consumers get pure JSON on stdout.

Agent Feedback

When you (or the agent) notice something off about this CLI, record it:

ollama-cloud-pp-cli feedback "the --since flag is inclusive but docs say exclusive"
ollama-cloud-pp-cli feedback --stdin < notes.txt
ollama-cloud-pp-cli feedback list --json --limit 10

Entries are stored locally at ~/.ollama-cloud-pp-cli/feedback.jsonl. They are never POSTed unless OLLAMA_CLOUD_FEEDBACK_ENDPOINT is set AND either --send is passed or OLLAMA_CLOUD_FEEDBACK_AUTO_SEND=true. Default behavior is local-only.

Write what surprised you, not a bug report. Short, specific, one line: that is the part that compounds.

Output Delivery

Every command accepts --deliver <sink>. The output goes to the named sink in addition to (or instead of) stdout, so agents can route command results without hand-piping. Three sinks are supported:

Sink Effect
stdout Default; write to stdout only
file:<path> Atomically write output to <path> (tmp + rename)
webhook:<url> POST the output body to the URL (application/json or application/x-ndjson when --compact)

Unknown schemes are refused with a structured error naming the supported set. Webhook failures return non-zero and log the URL + HTTP status on stderr.

Named Profiles

A profile is a saved set of flag values, reused across invocations. Use it when a scheduled agent calls the same command every run with the same configuration - HeyGen's "Beacon" pattern.

ollama-cloud-pp-cli profile save briefing --json
ollama-cloud-pp-cli --profile briefing chat chat --model example-value
ollama-cloud-pp-cli profile list --json
ollama-cloud-pp-cli profile show briefing
ollama-cloud-pp-cli profile delete briefing --yes

Explicit flags always win over profile values; profile values win over defaults. agent-context lists all available profiles under available_profiles so introspecting agents discover them at runtime.

Exit Codes

Code Meaning
0 Success
2 Usage error (wrong arguments)
3 Resource not found
4 Authentication required
5 API error (upstream issue)
7 Rate limited (wait and retry)
10 Config error

Argument Parsing

Parse $ARGUMENTS:

  1. Empty, help, or --help → show ollama-cloud-pp-cli --help output
  2. Starts with install → ends with mcp → MCP installation; otherwise → see Prerequisites above
  3. Anything else → Direct Use (execute as CLI command with --agent)

MCP Server Installation

Install the MCP binary from this CLI's published public-library entry or pre-built release, then register it:

claude mcp add ollama-cloud-pp-mcp -- ollama-cloud-pp-mcp

Verify: claude mcp list

Direct Use

  1. Check if installed: which ollama-cloud-pp-cli If not found, offer to install (see Prerequisites at the top of this skill).
  2. Match the user query to the best command from the Unique Capabilities and Command Reference above.
  3. Execute with the --agent flag: 
    ollama-cloud-pp-cli <command> [subcommand] [args] --agent
  4. If ambiguous, drill into subcommand help: ollama-cloud-pp-cli <command> --help.