--- name: pp-bestfoodtrucks description: "The only Best Food Trucks client that can tell you where a truck goes, not just what's parked outside today. Trigger phrases: `what food truck is at my office today`, `check the food truck schedule`, `when does my favorite food truck come back`, `best food trucks in Los Angeles`, `food truck menu and prices`, `use bestfoodtrucks`, `run bestfoodtrucks`." author: "Allen Lew" license: "Apache-2.0" argument-hint: " [args] | install cli|mcp" allowed-tools: "Read Bash" metadata: openclaw: requires: bins: - bestfoodtrucks-pp-cli install: - kind: go bins: [bestfoodtrucks-pp-cli] module: github.com/mvanhorn/printing-press-library/library/food-and-dining/bestfoodtrucks/cmd/bestfoodtrucks-pp-cli --- # Best Food Trucks — Printing Press CLI ## Prerequisites: Install the CLI This skill drives the `bestfoodtrucks-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 `$HOME/.local/bin` on macOS/Linux and `%LOCALAPPDATA%\Programs\PrintingPress\bin` on Windows: ```bash npx -y @mvanhorn/printing-press-library install bestfoodtrucks --cli-only ``` 2. Verify: `bestfoodtrucks-pp-cli --version` 3. Ensure the reported install directory is on `$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 1.26.6 or newer). This installs into `$GOPATH/bin` (default `$HOME/go/bin`), so add that directory to `$PATH` instead: ```bash go install github.com/mvanhorn/printing-press-library/library/food-and-dining/bestfoodtrucks/cmd/bestfoodtrucks-pp-cli@latest ``` If `--version` reports "command not found" after install, the runtime cannot see the binary directory on `$PATH`. Do not proceed with skill commands until verification succeeds. Best Food Trucks has no public API, no SDK, and no way to ask 'where else does this truck park' or 'summarize this week for my team.' This CLI talks directly to the same GraphQL backend the website and mobile apps use, then adds schedule digests, cuisine search, truck-centric reverse lookup, and cross-lot views the live site was never built to answer. ## When to Use This CLI Use this CLI for anonymous, read-only lookups of Best Food Trucks lot schedules, truck rotations, and menu data — checking what's at a specific office campus or lot, tracking a favorite truck's schedule across lots, or summarizing a week's schedule for a team. It is the right tool whenever a task needs structured or human-readable schedule/menu data faster than clicking through the bot-protected website by hand. ## Anti-triggers Do not use this CLI for: - Do not use this CLI to place real food orders or complete checkout — order-ahead requires a customer login and payment method, which this CLI does not automate. - Do not use this CLI to manage a customer's 'subscribe to lot schedule' notification preferences — that also requires a logged-in customer session not exercised by this build. - Do not use this CLI as a general nationwide food-truck search engine beyond lots/trucks/markets already known to Best Food Trucks — it only knows what Best Food Trucks itself tracks. ## Unique Capabilities These capabilities aren't available in any other tool for this API. ### Cross-schedule synthesis - **`lot digest`** — Turns a lot's upcoming schedule into ready-to-paste announcement text instead of raw structured data. _Reach for this when the task is 'summarize this week's schedule for humans,' not 'give me structured data.'_ ```bash bestfoodtrucks-pp-cli lot digest playa-district ``` - **`trucks find`** — Finds every upcoming shift at a lot matching a cuisine, without opening each shift page one at a time. _Use this when a user names a cuisine and a lot rather than a specific date — it walks the whole visible schedule window for you._ ```bash bestfoodtrucks-pp-cli trucks find --cuisine Thai --lot playa-district --json ``` - **`lots digest`** — Combines multiple lots' schedules into one view in a single command instead of visiting each lot's page separately. _Use this when a user tracks more than one regular lot (e.g., office campus plus a nearby favorite) and wants one combined answer._ ```bash bestfoodtrucks-pp-cli lots digest --lots playa-district,at-t-los-angeles --json ``` ### Truck-centric views the website never built - **`truck schedule`** — Shows every lot a specific truck visits, past and future — a view the Best Food Trucks website itself never built. _Use this to answer 'when does my favorite truck come back' or 'what other lots does this truck serve' — impossible from the live site's own navigation._ ```bash bestfoodtrucks-pp-cli truck schedule 11869 --json --select locations.records.startTime,locations.records.lot.name ``` - **`market hotlist`** — Ranks trucks operating in a city by review signal, a cross-truck aggregate the site never computes. _Use this for 'what's the best-rated truck in this city' rather than checking trucks one at a time._ ```bash bestfoodtrucks-pp-cli market hotlist los-angeles --limit 10 ``` ## Discovery Signals This CLI was generated with browser-observed traffic context. - Capture coverage: 25 API entries from 196 total network entries - Protocols: graphql (92% confidence), graphql_persisted_query (90% confidence), rest_json (75% confidence) - Generation hints: browser_http_transport, graphql_persisted_query, requires_protected_client, weak_schema_confidence - Candidate command ideas: create_graphql — Derived from observed POST /graphql traffic.; create_track_referrer.json — Derived from observed POST /track-referrer.json traffic.; create_track_request — Derived from observed POST /api/v1/intent_pixel/track_request traffic.; head_schedule.json — Derived from observed HEAD /_next/data/-dh7Qe2PYMUxWzoOQbyo3/lots/playa-district/schedule.json traffic.; list_179609_playa_district_on_2026_08_26.json — Derived from observed GET /_next/data/-dh7Qe2PYMUxWzoOQbyo3/shifts/179609-playa-district-on-2026-08-26.json traffic.; list_attribution_trigger — Derived from observed GET /attribution_trigger traffic.; list_can_track_visitor — Derived from observed GET /api/v1/intent_pixel/can_track_visitor traffic.; list_j — Derived from observed GET /j traffic. - Caveats: empty_payload: API-looking request returned an empty or null payload; schema confidence is weak.; empty_payload: API-looking request returned an empty or null payload; schema confidence is weak.; empty_payload: API-looking request returned an empty or null payload; schema confidence is weak.; empty_payload: API-looking request returned an empty or null payload; schema confidence is weak.; empty_payload: API-looking request returned an empty or null payload; schema confidence is weak.; empty_payload: API-looking request returned an empty or null payload; schema confidence is weak.; empty_payload: API-looking request returned an empty or null payload; schema confidence is weak.; empty_payload: API-looking request returned an empty or null payload; schema confidence is weak.; empty_payload: API-looking request returned an empty or null payload; schema confidence is weak.; empty_payload: API-looking request returned an empty or null payload; schema confidence is weak. ## Command Reference **graphql** — Raw GraphQL passthrough against the Best Food Trucks API. This is a generic escape-hatch endpoint; Phase 3 hand-writes typed, well-named commands (lot get/schedule/digest, shift get, truck schedule, market list/hotlist, trucks find, lots digest) on top of a hand-authored GraphQL client rather than relying on this endpoint's auto-emitted command surface. The CLI always sends full query text (no extensions.persistedQuery hash) to avoid depending on the server's Apollo Automatic Persisted Query cache. - `bestfoodtrucks-pp-cli graphql` — Execute a raw, read-only GraphQL query against the Best Food Trucks API (advanced escape hatch; mutations are out of scope) ### Finding the right command When you know what you want to do but not which command does it, ask the CLI directly: ```bash bestfoodtrucks-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 ### What's at my office today ```bash bestfoodtrucks-pp-cli lot schedule playa-district --json --select locationSchedule.dateAlias,locationSchedule.locations.truck.name ``` Pulls just today/tomorrow's truck name from the full schedule payload, instead of parsing the whole nested response by hand. ### Paste this week's schedule into Slack ```bash bestfoodtrucks-pp-cli lot digest playa-district ``` Produces the announcement text directly — no manual copy-paste from the website required. ### Find every Thai truck coming to my lot ```bash bestfoodtrucks-pp-cli trucks find --cuisine Thai --lot playa-district --json ``` Walks the visible schedule window and filters by cuisine tag, something no single API call does server-side. ### Track a favorite truck across every lot it visits ```bash bestfoodtrucks-pp-cli truck schedule 11869 --agent --select locations.records.startTime,locations.records.lot.name ``` Reverse lookup with --select narrows a potentially large history-plus-future list down to just the two fields an agent needs to answer 'when and where.' ### Combine two lots you care about into one view ```bash bestfoodtrucks-pp-cli lots digest --lots playa-district,at-t-los-angeles ``` One command instead of visiting two separate lot pages and manually merging the results. ## Auth Setup No authentication required. Run `bestfoodtrucks-pp-cli doctor` to verify setup. ## Agent Mode Add `--agent` to any command. Expands to: `--json --compact --no-input --no-color`. - **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 bestfoodtrucks-pp-cli graphql --query example-value --agent ``` - **Previewable** — `--dry-run` shows the request without sending - **Non-interactive** — never prompts, every input is a flag - **Read-only** — do not use this CLI for create, update, delete, publish, comment, upvote, invite, order, send, or other mutating requests ## Paths and state Agents should treat the CLI's path resolver as part of the runtime contract: - Use `--home ` for one invocation, or set `BESTFOODTRUCKS_HOME=` to relocate all four path kinds under one root. - Use per-kind env vars only when a specific kind must diverge: `BESTFOODTRUCKS_CONFIG_DIR`, `BESTFOODTRUCKS_DATA_DIR`, `BESTFOODTRUCKS_STATE_DIR`, `BESTFOODTRUCKS_CACHE_DIR`. - Resolution order is per-kind env var, `--home`, `BESTFOODTRUCKS_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 `data.db` and the local learning store. `state` contains persisted queries, jobs, and `teach.log`. `cache` contains regenerable HTTP/cache files. - This CLI stores no Best Food Trucks API credentials — the read surface is fully anonymous, so there is no `credentials.toml` or auth sidecar to manage. - Run `bestfoodtrucks-pp-cli doctor --fail-on warn` to surface path and connectivity 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": { "bestfoodtrucks": { "command": "bestfoodtrucks-pp-mcp", "env": { "BESTFOODTRUCKS_HOME": "/srv/bestfoodtrucks" } } } } ``` Fleet precedence: an inherited per-kind env var overrides an explicit `--home` for that kind. Use `BESTFOODTRUCKS_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 `BESTFOODTRUCKS_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 bestfoodtrucks-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": ["", "bestfoodtrucks-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 `bestfoodtrucks-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; `bestfoodtrucks-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 Cardinals + St. Louis Cardinals). Surface the ambiguity from context before committing to a resource. - `candidates_present` (top-level): the envelope carries a `candidates` section. Handle it via the candidates branch in Step 2 before anything else. - Top-level `no_learnings_for_query_family`: the table had no rows above the Jaccard floor. Pure cold start. ### Step 4: `teach &` after finalizing your response - always Teaching is unconditional. After resolving a query the store could not answer, background-teach the final resource mapping - no call-count threshold, no judging whether it was "worth" learning. The teach is the anchor of the loop: it triggers playbook synthesis for a family without a playbook, and same-referent phrasings fold into one family so near-duplicate teaches do no