--- name: pp-openai-ads description: "The first OpenAI Ads client that can write, and the only one that keeps local history. Trigger phrases: `check my ChatGPT ads`, `how is my openai ads campaign pacing`, `what changed in my ad account`, `which ads are getting tired`, `audit my openai ads account structure`, `use openai-ads`, `run openai-ads`." author: "bobe" license: "Apache-2.0" argument-hint: " [args] | install cli|mcp" allowed-tools: "Read Bash" metadata: openclaw: requires: bins: - openai-ads-pp-cli install: - kind: go bins: [openai-ads-pp-cli] module: github.com/mvanhorn/printing-press-library/library/marketing/openai-ads/cmd/openai-ads-pp-cli --- # OpenAI Ads — Printing Press CLI ## Prerequisites: Install the CLI This skill drives the `openai-ads-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 `$HOME/.local/bin` on macOS/Linux and `%LOCALAPPDATA%\Programs\PrintingPress\bin` on Windows: ```bash npx -y @mvanhorn/printing-press-library install openai-ads --cli-only ``` 2. Verify: `openai-ads-pp-cli --version` 3. Ensure the reported install directory is on `$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 1.26.5 or newer). This installs into `$GOPATH/bin` (default `$HOME/go/bin`), so add that directory to `$PATH` instead: ```bash go install github.com/mvanhorn/printing-press-library/library/marketing/openai-ads/cmd/openai-ads-pp-cli@latest ``` If `--version` reports "command not found" after install, the runtime cannot see the binary directory on `$PATH`. Do not proceed with skill commands until verification succeeds. Covers the whole Advertiser API surface rather than a read-only slice of it, and mirrors your account into SQLite so questions the REST API structurally cannot answer become one command. Pacing, drift, creative fatigue, and structural audits all come from local snapshots. Every monetary value is rendered in your account currency instead of raw micros. ## When to Use This CLI Reach for this when working with ChatGPT Ads campaigns programmatically: auditing account structure, checking spend pacing, diffing what changed, or creating and updating campaigns, ad groups, and ads. It is strongest when the question spans more than one resource or more than one point in time, because the local mirror can join and compare where the REST API cannot. ## Anti-triggers Do not use this CLI for: - Do not use this CLI for the OpenAI Platform API; model, file, and batch calls belong to api.openai.com and a different credential. - Do not use this CLI to send server-side conversion events; that is a separate ingestion endpoint with its own event-scoped key. - Do not use this CLI to complete business verification or brand review, which are console-only workflows. - Do not expect local commands to reflect changes made in Ads Manager until after a sync. ## Unique Capabilities These capabilities aren't available in any other tool for this API. ### Local history the API does not keep - **`pace`** — See whether a campaign will underspend or blow through its cap before the period ends. _Reach for this instead of raw insights when the question is about trajectory rather than a current number._ ```bash openai-ads-pp-cli pace --agent ``` - **`drift`** — Show what changed across campaigns, ad groups, and ads between two syncs. _Use this to answer 'what changed and when', which no single API call can report._ ```bash openai-ads-pp-cli drift --since 7d --agent ``` - **`fatigue`** — Rank ads by click-through decay so tired creative is obvious before spend is wasted. _Pick this over ad insights when the question is whether performance is declining, not what it is today._ ```bash openai-ads-pp-cli fatigue --limit 10 --agent ``` - **`review-watch`** — Track approval and review status transitions across the account and every ad. _Use this to catch a flip to rejected or in_review that a status read would not reveal as a change._ ```bash openai-ads-pp-cli review-watch --agent ``` ### Cross-entity joins the API cannot do - **`bid-check`** — Flag ad groups whose maximum bid is irrational against the parent campaign budget. _Catches configurations that permit only a handful of clicks per day before any spend happens._ ```bash openai-ads-pp-cli bid-check --agent ``` - **`orphans`** — Find ad groups with no ads, campaigns with no delivery, and audiences nothing references. _Use this for structural dead weight rather than performance questions._ ```bash openai-ads-pp-cli orphans --agent ``` - **`tree`** — Render the whole campaign, ad group, and ad hierarchy with status, budget, and review state. _Start here to orient in an unfamiliar account before drilling into any single resource._ ```bash openai-ads-pp-cli tree --agent ``` ### Readability - **`geo resolve`** — Turn the bare location IDs in campaign targeting into readable place names. _Use this whenever targeting output shows numeric IDs you cannot interpret._ ```bash openai-ads-pp-cli geo resolve --agent ``` ## Command Reference **ad-account** — Manage ad account - `openai-ads-pp-cli ad-account activate-method` — Activate the ad account. - `openai-ads-pp-cli ad-account get-insights-method` — Get ad account insights aggregated by time granularity. - `openai-ads-pp-cli ad-account get-method` — Get metadata for the ad account - `openai-ads-pp-cli ad-account pause-method` — Pause the ad account. - `openai-ads-pp-cli ad-account update-method` — Update ad account brand metadata. **ad-groups** — Manage ad groups - `openai-ads-pp-cli ad-groups create-method` — Create an ad group for a campaign - `openai-ads-pp-cli ad-groups get-method` — Get an ad group - `openai-ads-pp-cli ad-groups list-method` — Get all ad groups for a campaign - `openai-ads-pp-cli ad-groups update-method` — Update an ad group **ads** — Manage ads - `openai-ads-pp-cli ads create-method` — Create an ad for an ad group - `openai-ads-pp-cli ads get-method` — Get an ad - `openai-ads-pp-cli ads list-method` — Get all ads for an ad group - `openai-ads-pp-cli ads update-method` — Update an ad **campaigns** — Manage campaigns - `openai-ads-pp-cli campaigns create-method` — Create a campaign for an ad account - `openai-ads-pp-cli campaigns get-method` — Get a campaign - `openai-ads-pp-cli campaigns list-method` — Get all campaigns for an ad account - `openai-ads-pp-cli campaigns update-method` — Update a campaign **conversions** — Manage conversions - `openai-ads-pp-cli conversions create-api-key-method` — Create a Conversions API key for the currently authenticated ad account. - `openai-ads-pp-cli conversions create-event-setting-method` — Create a conversion event setting for the currently authenticated ad account. - `openai-ads-pp-cli conversions create-source-method` — Create a conversion pixel. - `openai-ads-pp-cli conversions list-event-settings-method` — List conversion event settings for the currently authenticated ad account. - `openai-ads-pp-cli conversions post-insights-method` — Get attributed conversion totals for the authenticated ad account. **custom-audiences** — Manage custom audiences - `openai-ads-pp-cli custom-audiences create-method` — Create a custom audience for the authenticated ad account. - `openai-ads-pp-cli custom-audiences create-upload-method` — Create a custom audience from an uploaded file and start processing. - `openai-ads-pp-cli custom-audiences get-method` — Get a custom audience for the authenticated ad account. - `openai-ads-pp-cli custom-audiences list-method` — List custom audiences for the authenticated ad account. **geo-lookup** — Manage geo lookup - `openai-ads-pp-cli geo-lookup` — Search DMA and standard region codes for advertiser geo targeting. **upload** — Manage upload - `openai-ads-pp-cli upload` — Upload an image URL or image file and return a file id ### Finding the right command When you know what you want to do but not which command does it, ask the CLI directly: ```bash openai-ads-pp-cli which "" ``` `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 ### Orient in an unfamiliar account ```bash openai-ads-pp-cli tree --agent ``` Renders the whole hierarchy with status and budget so you can see the shape before touching anything. ### Narrow a verbose campaign payload ```bash openai-ads-pp-cli campaigns list --agent --select data.id,data.name,data.status,data.budget.daily_spend_limit_micros ``` Campaign objects carry targeting and landing page blocks; selecting fields keeps agent context small. ### Find tired creative ```bash openai-ads-pp-cli fatigue --limit 5 --agent ``` Ranks ads by click-through decay across stored snapshots rather than a single reading. ### See what changed this week ```bash openai-ads-pp-cli drift --since 7d --agent ``` Diffs local snapshots to surface status, budget, bid, and creative changes the API keeps no record of. ### Preview a mutation safely ```bash openai-ads-pp-cli campaigns pause campaign-method cmpn_example --dry-run ``` Shows the request that would be sent without changing anything in the live account. ## Auth Setup Ads Manager issues two different keys and they are easy to confuse. The Ads API key comes from the Settings tab and is scoped to one ad account; set it as OPENAI_ADS_API_KEY. The Conversions API key comes from the Conversions tab, is scoped to a single conversion event, and returns 403 Unauthorized to read ads data if used for ads calls. Set that one as OPENAI_ADS_CONVERSIONS_API_KEY only if you send server-side conversion events. Run doctor to see which key is configured. Run `openai-ads-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: ```bash openai-ads-pp-cli ad-account activate-method --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: ```json { "meta": {"source": "live" | "local", "synced_at": "...", "reason": "..."}, "results": } ``` 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 AND no machine-format flag (`--json`, `--csv`, `--compact`, `--quiet`, `--plain`, `--select`) is set — piped/agent consumers and explicit-format runs get pure JSON on stdout. ## Paths and state Agents should treat the CLI's path resolver as part of the runtime contract: - Use `--home ` for one invocation, or set `OPENAI_ADS_HOME=` to relocate all four path kinds under one root. - Use per-kind env vars only when a specific kind must diverge: `OPENAI_ADS_CONFIG_DIR`, `OPENAI_ADS_DATA_DIR`, `OPENAI_ADS_STATE_DIR`, `OPENAI_ADS_CACHE_DIR`. - Resolution order is per-kind env var, `--home`, `OPENAI_ADS_HOME`, XDG (`XDG_CONFIG_HOME`, `XDG_DATA_HOME`, `XDG_STATE_HOME`, `XDG_CACHE_HOME`), then platform defaults. - `config` contains settings like `config.toml` and profiles. `data` contains `credentials.toml`, `data.db`, cookies, and auth sidecars. `state` contains persisted queries, jobs, and `teach.log`. `cache` contains regenerable HTTP/cache files. - Stored secrets live in `credentials.toml` under the data dir. Existing legacy `config.toml` secrets are read for compatibility and leave `config.toml` on the first auth write. - Run `openai-ads-pp-cli doctor --fail-on warn` to surface path and credential-location warnings. `agent-context` exposes a schema v4 `paths` block for agents that need the resolved dirs. - For MCP, pass relocation through the MCP host config. The MCP binary does not inherit CLI flags: ```json { "mcpServers": { "openai-ads": { "command": "openai-ads-pp-mcp", "env": { "OPENAI_ADS_HOME": "/srv/openai-ads" } } } } ``` Fleet precedence: an inherited per-kind env var overrides an explicit `--home` for that kind. Use `OPENAI_ADS_HOME` or per-kind vars as durable fleet levers, and use `--home` only for a single invocation. Relocation is not reversible by unsetting env vars; move files manually before clearing `OPENAI_ADS_HOME`, or `doctor` will not find credentials left under the former root. ## Automatic learning This CLI ships a self-capturing learning loop. The CLI does its own bookkeeping: every invocation is journaled locally, a failed flag followed by a corrected retry auto-derives a `flag_alias` candidate, and a `teach` on a query family without a playbook auto-synthesizes a `playbook_candidate` from the session's journal. Your job is judgment only: `recall` first, act on surfaced candidates, `teach` the final answer, `playbook amend` when you observe a correction. You never record failures by hand. ### Step 1: `recall` before any discovery Before list/search/drill commands on a new user question, run: ```bash openai-ads-pp-cli recall "" --agent ``` The response envelope: ```json { "query": "...", "normalized": "", "query_entities": ["..."], "found": true | false, "match_score": 0.0, "results": [ { "resource_id": "...", "resource_type": "...", "venue": "...", "confidence": 2, "entity_match": "exact|partial|unknown", "source": "taught|preseed|pattern", "warnings": ["..."] } ], "mismatches": [ /* only when --debug-mismatches */ ], "warnings": [ /* top-level */ ], "candidates": [ { "id": 12, "class": "flag_alias | playbook_candidate", "summary": "...", "sightings": 3, "last_seen": "...", "rationale": "...", "next_action": ["", "openai-ads-pp-cli learnings confirm 12"] } ], "playbook": { "query_family": "...", "playbook": { "steps": [ { "cmd": "", "purpose": "..." } ], "entity_slots": ["$ENTITY"], "expected_tool_calls": 3 }, "slots_resolved": { "$ENTITY": { "token": "", "canonical": "" } }, "notes": "" }, "notes": "" } ``` Empty-store short-circuit: if the store has no learnings, playbooks, or candidates yet (recall finds nothing and `learnings list` and `learnings candidates` are both empty), skip recall for the rest of this session instead of taxing every query; resume recall-first once something has been taught. ### Step 2: decision tree Read `candidates`, `playbook`, `notes`, `results[0]`, and warnings in that order: ``` if Candidates present (warnings include "candidates_present"): -> candidates are try-then-confirm, never facts. Follow each candidate's two-step next_action verbatim: run the trial command first, then run `learnings confirm ` only after the trial verified the behavior. Reject a wrong candidate with `learnings reject `. -> NEVER re-teach something recall surfaced as a candidate; confirm or reject that candidate instead of teaching a duplicate. -> candidates ride alongside playbooks and resource hits, not instead of them; continue with the branches below after acting on them. if Playbook present: -> READ Playbook.notes verbatim FIRST (workarounds + gotchas the CLI surface doesn't expose) -> replay Playbook.steps in order, substituting Playbook.slots_resolved entries for the entity slot tokens. If a step's slot is unresolved, fall back to discovery for that step only. -> the Playbook's expected_tool_calls is a budget; if you find yourself running materially more, record the divergence via `openai-ads-pp-cli playbook amend` at end-of-session. elif Notes present (no Playbook): -> read Notes verbatim before any discovery step; they carry known gotchas for this query family even when no structured choreography exists yet. elif Found AND Results[0].EntityMatch == "exact" AND Results[0].Confidence >= 2: -> skip discovery; fetch live data for Results[*].ResourceID in parallel elif Found AND Results[0].EntityMatch == "partial": -> candidate hint, NOT a hit; read the resource title to validate before trusting elif (any row in Mismatches[] when --debug-mismatches was passed): -> treat as cold start; the stored learning is for a different entity (different canonical resolved from query_entities) else: // Found == false, no playbook, no notes -> cold start; run discovery normally; teach the answer afterward (Step 4). If the family has no playbook yet, that teach auto-synthesizes a playbook candidate from this session's journal - you do not need to record one by hand. ``` Playbook and Notes are orthogonal to the per-resource path. A recall response can carry both a Playbook AND a `Results[]` hit - use both: the Playbook tells you which choreography to run; the resource hits short-circuit specific steps. Default to skipping `mismatches`; pass `--debug-mismatches` only when investigating cold-start surprises. Candidate judgment details: `learnings confirm ` prints the candidate's full payload before materializing it - check that the printed payload matches the behavior you verified. `learnings reject ` tombstones the derivation signature so the same candidate does not resurface. The envelope carries only the few candidates worth acting on now; `openai-ads-pp-cli learnings candidates` lists the full open set. Graceful degradation: if `learnings confirm` is an unknown command, you are driving an older binary - ignore the candidates guidance and follow the rest of the protocol. ### Step 3: always read `warnings` - `low_confidence`: row exists at `confidence<2`. Treat as a hint, not a skip-discovery hit. - `resource_not_in_store`: the local store doesn't have the resource the learning points at. The match validator couldn't classify entities — direct-fetch and re-evaluate. - `cross_alias_match` (per-result): the row was taught under a different alias and matched the live query's canonical via `entity_lookups` (e.g., a "USA" teach satisfying a "United States" recall). Trust the resource_id. - `similar_shape_different_entity:` (top-level): a structurally matching row exists but its canonical entity differs from the live query's. Treated as cold start; the warning carries the conflicting canonical as a hint, but the row is NOT promoted into Results. - `ambiguous_alias` (top-level): a single query entity resolved to multiple canonicals (e.g., "Cards" → Arizona Card