Agent skill

Pp Forkable

by mvanhorn in mvanhorn/printing-press-library

A CLI and MCP server for your Forkable office-lunch program, with a local database and history, spend, and preference queries the web app cannot answer, plus commands to set, confirm, and skip meal…

Apache-2.0Auto-check: notesAgent Workflows

Install Pp Forkable

skills CLI
$ npx skills add mvanhorn/printing-press-library --skill pp-forkable -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install mvanhorn/printing-press-library pp-forkable --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/mvanhorn/printing-press-library.git skills-src && mkdir -p .claude/skills && cp -r skills-src/cli-skills/pp-forkable .claude/skills/pp-forkable && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
pp-forkable
GitHub stars
2.1k
Token cost
~9.7k tokens
SKILL.md length
4,222 words
Files
1
Skills in repo
505
Repo updated
First seen
Licence
Apache-2.0

At a glance

A CLI and MCP server for your Forkable office-lunch program, with a local database and history, spend, and preference queries the web app cannot answer, plus commands to set, confirm, and skip meal…

  • Works in 6 steps: recall before any discovery → decision tree → always read warnings → …
  • Phrases: what have I eaten on forkable
  • SKILL.md covers Prerequisites: Install the CLI, When to Use This CLI, Anti-triggers and Unique Capabilities, plus 6 more sections
  • Calls go, claude and npx

What it does

Pp Forkable is an agent skill from mvanhorn/printing-press-library. A CLI and MCP server for your Forkable office-lunch program, with a local database and history, spend, and preference queries the web app cannot answer, plus commands to set, confirm, and skip meal orders (dry-run by default; --confirm to apply). Trigger phrases: what have I eaten on forkable, forkable lunch spend this month, which forkable venues do we use most, did my forkable meals match my dietary preferences, forkable week ahead, set my forkable meal, skip forkable tomorrow, use forkable, run forkable.

Its SKILL.md is about 9.7k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Agent Workflows. It works with Model Context Protocol. The repository describes itself as: Official library of CLIs generated by the CLI Printing Press. Endorsed, tested, and community-contributed. The licence is Apache-2.0.

When your agent uses it

  • Phrases: what have I eaten on forkable
  • Forkable lunch spend this month
  • Which forkable venues do we use most
  • Did my forkable meals match my dietary preferences

Example prompts

  • “/pp-forkable”

Requirements

  • Node.js
  • Pre-approved tools (allowed-tools): Read, Bash

Workflow steps

6 steps, taken from the step headings in SKILL.md.

  1. recall before any discovery
  2. decision tree
  3. always read warnings
  4. teach & after finalizing your response - always
  5. playbooks - optional flags, automatic synthesis
  6. playbook amend & when your debug response identifies a correction

What it can do on your machine

Read from SKILL.md and the folder at commit 7638ad4. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves these tools, so the agent can use them without asking each time:

    • Read
    • Bash

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • go
    • claude
    • npx

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use npx, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Pp Forkable loads about 9.7k tokens when it runs. Until then it costs about 136 tokens; SKILL.md has 4,222 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~136
When it runs · the whole SKILL.md, loaded when a task matches
~9.7k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check: notes

The automated check noted patterns worth knowing about, such as sudo or a known installer.

  • NotePre-approves every shell command (allowed-tools: Bash)SKILL.md
    allowed-tools: Read, Bash

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from mvanhorn/printing-press-library at commit 7638ad4, republished under its Apache-2.0 licence (© mvanhorn). 4,222 words, ~9,680 tokens.

Download SKILL.mdSave it as .claude/skills/pp-forkable/SKILL.md (or your agent's skills folder).
name
pp-forkable
description
A CLI and MCP server for your Forkable office-lunch program, with a local database and history, spend, and preference queries the web app cannot answer, plus commands to set, confirm, and skip meal orders (dry-run by default; --confirm to apply). Trigger phrases: `what have I eaten on forkable`, `forkable lunch spend this month`, `which forkable venues do we use most`, `did my forkable meals match my dietary preferences`, `forkable week ahead`, `set my forkable meal`, `skip forkable tomorrow`, `use forkable`, `run forkable`.
allowed-tools
Read, Bash
author
Allen Lew
license
Apache-2.0
argument-hint
<command> [args] | install cli|mcp
<!-- GENERATED FILE — DO NOT EDIT.
     This file is a verbatim mirror of library/food-and-dining/forkable/SKILL.md,
     regenerated post-merge by tools/generate-skills/. Hand-edits here are
     silently overwritten on the next regen. Edit the library/ source instead.
     See the repository agent guide, section "Generated artifacts: registry.json, cli-skills/". -->

Forkable — Printing Press CLI

Prerequisites: Install the CLI

This skill drives the forkable-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 forkable --cli-only
  2. Verify: forkable-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/forkable/cmd/forkable-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.

Forkable exposes no public API. This CLI reverse-engineers the my-account app's GraphQL surface into a Go binary with clean read commands and agent-native output. On top of the raw reads it adds longitudinal views the product never shows — served-meal history, preference-vs-served drift, spend trends, allowance utilization, venue rotation, and a week-ahead digest — all fetched live from Forkable. It also exposes the my-account app's own meal-management mutations as meal set, meal set-all, meal confirm, meal skip, and reorder; these are dry-run by default and only place, confirm, or skip real orders when you pass --confirm. A fifth meal command, meal feedback, reports a problem with a delivered meal to Forkable via whichever of two real mechanisms actually handles it (both discovered by static analysis of the app's JS bundle, not documented anywhere): addMemberReportedIssue, a real GraphQL mutation for missing-item/missing-side reports with refund/credit resolution tracking and a same-day cutoff, or the REST "Contact Support" endpoint for everything else; see the Command Reference below.

When to Use This CLI

Use this CLI when an agent or script needs to query a Forkable office-lunch program from the terminal: what meals were served, how much was spent, which venues rotate, and whether served meals match stated dietary preferences. It is the only programmatic surface for Forkable and the only way to get longitudinal history the web app does not aggregate. It can also set, confirm, and skip upcoming meal deliveries on the user's behalf — those commands are dry-run by default and require --confirm to apply. All commands fetch live from Forkable (no local sync step).

Anti-triggers

Do not use this CLI for:

  • Do not place, change, or cancel meal orders without explicit user intent — the meal set, meal set-all, meal confirm, meal skip, and reorder commands mutate the real account and require --confirm; run them dry-run first and confirm the printed mutation before adding --confirm.
  • Do not use this CLI to add or remove club members or manage billing — use the Forkable web app.
  • Do not use this CLI for real-time delivery tracking notifications — those come via Forkable's email/Slack/SMS.
  • Do not run meal feedback ... --confirm without explicit user intent — like the four mutation commands above, it is a real, one-way action: it files an actual report or support request with Forkable and cannot be un-sent. Preview it first (the default, no---confirm behavior) and confirm the printed request before adding --confirm.
  • Do not use --category missing-item/missing-side after Forkable's same-day cutoff (reportMissingItemCutoff, typically 1pm local restaurant time) — the mutation is server-side rejected past it. The CLI checks this proactively and fails fast with the cutoff time; fall back to --category quality/other to reach Forkable support through the always-available contact form instead.

Unique Capabilities

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

Local history that compounds
  • served-history — See every meal actually served to you over time, with date, venue, price, and dietary level.

    Reach for this when an agent needs a longitudinal view of what a person has eaten, not just the current delivery.

    bash
    forkable-pp-cli served-history --since 90d --agent
  • preference-drift — Flag served meals that violate your stated dislikes or dietary restrictions, or miss your likes.

    Use this to audit whether auto-selection is actually honoring dietary preferences over time.

    bash
    forkable-pp-cli preference-drift --since 60d --agent
  • venue-rotation — Rank venues by how often they've served you and how recently.

    Use this to spot venue fatigue or under-used favorites across the whole synced window.

    bash
    forkable-pp-cli venue-rotation --since 120d --agent
Making the opaque legible
  • why-picked — Explain why a delivery's meal was auto-selected by ranking candidate items and their scores.

    Pick this to explain a single day's auto-selected meal; use preference-drift for aggregate conformance.

    bash
    forkable-pp-cli why-picked --delivery 1219480 --agent
Finance and allowances
  • spend-trend — Bucket lunch spend into per-week or per-month totals with CSV export.

    Reach for this when finance needs a time series of lunch cost, not a single delivery receipt.

    bash
    forkable-pp-cli spend-trend --since 6mo --by month --csv
  • allowance-burn — Show granted-vs-consumed allowance utilization per club, including multi-club comparison.

    Use this to see which teams are over- or under-using their lunch budget.

    bash
    forkable-pp-cli allowance-burn --by club --csv
Agent-native plumbing
  • upcoming-digest — One agent-shaped line per upcoming day: date, venue, auto-selected item, price, allowance headroom.

    Pick this for a quick 'what's coming this week' summary an agent can read in one shot.

    bash
    forkable-pp-cli upcoming-digest --agent

Command Reference

account —

  • forkable-pp-cli account — Show the authenticated Forkable user: profile, roles, dietary preferences (likes/dislikes/restrictions), companies

buffet_addresses —

  • forkable-pp-cli buffet-addresses — List your buffet delivery addresses (street, city, postal code, coordinates, club).

clubs —

  • forkable-pp-cli clubs — List meal clubs (teams/offices) you belong to or manage, with delivery address, delivery days, allowances

csrf —

  • forkable-pp-cli csrf — Fetch a CSRF token. Read-only.

deliveries —

  • forkable-pp-cli deliveries in-progress-ids — List IDs of deliveries currently in progress.
  • forkable-pp-cli deliveries list — List your meal deliveries from a given date forward, including per-delivery orders, chosen menu items, receipts

meal_scores —

  • forkable-pp-cli meal-scores — Show meal auto-selection scores (menuId, itemId, score) for a delivery and user across candidate menus.

menus —

  • forkable-pp-cli menus — Get menu(s) with venue, sections, items, prices, dietary levels, ratings, and modifiers.

notifications —

  • forkable-pp-cli notifications — List account notifications shown in the my-account app (title, description, links, publish window).

venue_usage —

  • forkable-pp-cli venue-usage — Get per-venue usage keyed by venue id over a date range. Requires venue ids and from/to dates inlined in the query.

Stale default dates on deliveries list and venue-usage. These two raw GraphQL passthrough commands ship a default --query with an example date baked directly into the query text. Running either with no flags can silently return an empty or misleading result (e.g. deliveries list with no overrides returns {"count":0} even when real deliveries exist) rather than an error. Always override the date argument with a real date — today's date for forward-looking reads, or a far-past date (e.g. 2000-01-01, as served-history/allowance-burn already do internally) for full-history reads — before trusting a zero/empty result from either command. Other passthrough reads (menus, meal-scores) instead ship placeholder ids that need replacing — see each command's own entry above.

meal management (writes) — place real orders and spend; dry-run by default, pass --confirm to apply.

  • forkable-pp-cli meal set <deliveryId> --item <id> --menu <id> [--modifier <modifierId>:<optionId>] [--replace-piece <uuid>] [--note <text>] [--confirm] — Override the auto-picked meal for one delivery day (replacePiece). --replace-piece takes the current piece's UUID (from deliveries list).
  • forkable-pp-cli meal set-all --deliveries <id,id> --item <id> --menu <id> [--modifier <modifierId>:<optionId>] [--confirm] — Apply one meal item across several delivery days (replaceAllPieces).
  • forkable-pp-cli meal confirm <deliveryId> [--unconfirm] [--confirm] — Confirm (or --unconfirm) a delivery day (confirmDelivery).
  • forkable-pp-cli meal skip <deliveryId> [--confirm] — Skip / cancel one or more delivery days (removeDelivery); --deliveries <id,id> skips several.
  • forkable-pp-cli reorder <fromDate> --onto <deliveryId> [--replace-piece <uuid>] [--confirm] — Repeat the meal you had on a past date onto an upcoming delivery day.
  • forkable-pp-cli meal feedback <deliveryId> --category <missing-item|missing-side|wrong-item|quality|late|other> [--note <text>] [--piece <uuid>] [--item <name>] [--venue <name>] [--request-refund|--request-credit] [--confirm] — Report a problem with a delivered meal to Forkable, via whichever real mechanism actually handles the category (see below). Dry-run by default like the four mutations above.
  • forkable-pp-cli meal feedback list [--limit N] — Lists this CLI's own local record of past attempts. forkable-pp-cli meal feedback list --delivery <id> instead fetches Forkable's own myReportedIssues live for that delivery — the real ground truth for missing-item/missing-side reports (resolution, refund/credit status), bypassing the local ledger.

meal feedback uses two real mechanisms, picked by --category — neither is a GraphQL "mutation" in the everyday sense of the four above, and neither is documented anywhere.

  1. --category missing-item/missing-side → addMemberReportedIssue, a real GraphQL mutation (same mutation ($input: ...!) {...} wire shape as the four above) backing the my-account app's per-meal "Report Missing Item" widget. Discovered in the SPA bundle's module 372 (MealReportIssue component). Requires --piece (the piece/meal UUID from deliveries list/served-history); the order id is resolved automatically by looking up which order contains that piece. Ties into the courier's Onfleet delivery photos and supports --request-refund/--request-credit resolution tracking via a myReportedIssues field on the delivery. Time-gated: Forkable enforces a same-day cutoff (delivery.reportMissingItemCutoff, confirmed live as 1pm local restaurant time) — a submission after it is rejected server-side with a payload-level error. This CLI checks the cutoff proactively before attempting the mutation and fails fast with the exact time if it's already passed, rather than spending a live call on a doomed request.
  2. --category wrong-item/quality/late/other → POST /submit_contact_form, a plain REST call (not GraphQL) behind the generic "Contact Support" modal, with {name, email, subject, message, market_id}. Discovered in the SPA bundle's module 29822 (getSubjectOptions()/submitContactForm()) — its own copy for "Other Issues" says outright "please use Forkable Support", i.e. Forkable's own UI routes here too for anything the dedicated missing-item flow doesn't cover. --category maps to Forkable's subject list (wrong-item/quality→"Report Issue", late→"Delivery Status", other→"General Inquiry"); delivery id/venue/item/piece have no dedicated field on this form, so they're folded into the message text alongside --note. Not time-gated — always available, including as a fallback once the missing-item cutoff has passed.

Building an accurate preview for either mechanism touches the network even without --confirm (mechanism 1 resolves the order id from --piece; mechanism 2 resolves your name/email and the delivery's market id) — unlike the four mutations above, which preview fully offline. Every --confirm attempt (submitted or rejected by Forkable, either mechanism) is also appended to a local meal-feedback.jsonl ledger under the CLI's data dir — a preview-only run makes no live-writing attempt, so it is not logged. This ledger is separate from the top-level feedback command (which is for friction with this CLI, not with a meal).

--replace-piece is effectively always required for meal set and reorder. Forkable auto-selects a candidate meal for every delivery day before you ever touch it — in practice there is no delivery day with a genuinely empty slot, so omitting --replace-piece is a rare/theoretical path, not the default case. Dry-run mode does not validate this. A dry-run without --replace-piece renders a plausible-looking mutation preview and only fails at --confirm time, with a raw GraphQL error (oldPieceId ... Expected value to not be null). Before running --confirm, always look up the target delivery's current piece id via deliveries list (its orders[].pieces[].id) and pass it with --replace-piece. meal set-all has no --replace-piece flag — replaceAllPieces takes no old-piece id at all, so this caveat doesn't apply to it.

Required item options — --modifier. An item with a required option group (e.g. "Choose Protein", min: 1) is rejected by Forkable unless you send a selection. Pass --modifier <modifierId>:<optionId>[,<optionId>...] (repeatable) — the CLI builds the selectionsHash ({"16":[10]}) the replacePiece mutation needs. Find modifier and option ids under each item's modifiers in the menus output. Example: forkable-pp-cli meal set 12345 --item 678 --menu 90 --modifier 16:10 --replace-piece <uuid> --confirm.

Finding the right command

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

bash
forkable-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

What have I eaten this quarter
bash
forkable-pp-cli served-history --since 90d --agent --select date,venue,name,price

Longitudinal list of served meals, narrowed to the high-signal fields to keep agent context small.

Audit dietary conformance
bash
forkable-pp-cli preference-drift --since 60d --json

Flags any served meal that conflicts with your stated dislikes or restrictions.

Monthly lunch spend for finance
bash
forkable-pp-cli spend-trend --since 6mo --by month --csv

Per-month spend totals exported as CSV for a budget close.

Which teams are burning their allowance
bash
forkable-pp-cli allowance-burn --by club --csv

Granted-vs-consumed allowance utilization per club, side by side.

This week's lunch at a glance
bash
forkable-pp-cli upcoming-digest --agent

One compact line per upcoming day for a quick agent-readable briefing.

Flag a problem with today's meal
bash
forkable-pp-cli meal feedback 12345 --category missing-item --note "Rice was missing, that was pictured" --piece <uuid> --confirm
forkable-pp-cli meal feedback list --delivery 12345 --agent

Files a real "Missing Meal" report with Forkable before the same-day cutoff (dry-run without --confirm — preview it first), then checks its live resolution status. Find the deliveryId/--piece via upcoming-digest or deliveries list. Past the cutoff, use --category quality/other instead to reach Forkable support through the always-available contact form.

Auth Setup

Forkable authenticates with a browser session cookie plus a per-request CSRF token fetched from /api/v2/csrf_token. Log in to forkable.com in Chrome, then run 'forkable-pp-cli auth login --chrome' to import your session. There is no API key.

Run forkable-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
    forkable-pp-cli buffet-addresses --query example-value --agent --select id,name,status
  • Previewable — --dry-run shows the request without sending

  • Safe writes — read commands query Forkable; the write commands (meal set, meal set-all, meal confirm, meal skip, reorder, meal feedback) are dry-run by default and only mutate the account (or, for meal feedback, submit a real support request) when invoked with --confirm

  • Live fetch — commands query Forkable directly over GraphQL; there is no local sync/cache step

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

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": <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 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 <dir> for one invocation, or set FORKABLE_HOME=<dir> to relocate all four path kinds under one root.

  • Use per-kind env vars only when a specific kind must diverge: FORKABLE_CONFIG_DIR, FORKABLE_DATA_DIR, FORKABLE_STATE_DIR, FORKABLE_CACHE_DIR.

  • Resolution order is per-kind env var, --home, FORKABLE_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 forkable-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": {
        "forkable": {
          "command": "forkable-pp-mcp",
          "env": {
            "FORKABLE_HOME": "/srv/forkable"
          }
        }
      }
    }

Fleet precedence: an inherited per-kind env var overrides an explicit --home for that kind. Use FORKABLE_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 FORKABLE_HOME, or doctor will not find credentials left under the former root.

Show full SKILL.md (1,762 more words)Show less

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
forkable-pp-cli recall "<user's question>" --agent

The response envelope:

json
{
  "query": "...",
  "normalized": "<normalized form>",
  "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": ["<trial command>", "forkable-pp-cli learnings confirm 12"] }
  ],
  "playbook": {
    "query_family": "...",
    "playbook": {
      "steps": [ { "cmd": "<command with {slot} substitution>", "purpose": "..." } ],
      "entity_slots": ["$ENTITY"],
      "expected_tool_calls": 3
    },
    "slots_resolved": { "$ENTITY": { "token": "<live token>", "canonical": "<canonical>" } },
    "notes": "<workarounds + gotchas for this query family>"
  },
  "notes": "<duplicate surface for non-playbook callers>"
}

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 <id>` only after the trial verified the behavior.
       Reject a wrong candidate with `learnings reject <id>`.
    -> 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 `forkable-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 <id> prints the candidate's full payload before materializing it - check that the printed payload matches the behavior you verified. learnings reject <id> tombstones the derivation signature so the same candidate does not resurface. The envelope carries only the few candidates worth acting on now; forkable-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:<canonical> (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.
  • lookup_refresh_available (top-level): an entity in the query has no lookup row yet. This CLI fetches live (no sync command); populate lookups by running the relevant read command (e.g. deliveries list, clubs list).
  • 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 not fragment the store. Fire it after assembling your user-facing response but BEFORE emitting it, with a shell & so the call returns immediately:

bash
forkable-pp-cli teach --query "<user's question>" --resource-type <type> --resource <id1> --resource <id2>
# (append shell `&` to background it)

Silent on success. Errors only land in teach.log under the resolved state dir. Teach the most specific resource - if the user asked a broad question and you walked through parent records to find the specific answer, teach the leaf id, not the parent. The CLI uses seeded entity_lookups for cross-alias resolution at recall time, so a teach under one alias (e.g., "Niners") satisfies future queries under another alias (e.g., "49ers", "San Francisco") automatically.

PII rule: teach the structural question with identifiers stripped - never include names, emails, phone numbers, account ids, or other personal identifiers in taught queries or notes. The CLI scans teach queries for obvious email/phone shapes and warns, but does not block; strip before teaching rather than relying on the warning.

Step 5: playbooks - optional flags, automatic synthesis

You do not need to decide whether a session "deserves" a playbook: a teach on a family without one auto-synthesizes a playbook_candidate from the session's journal, and the next session judges it via confirm/reject. Attach explicit playbook flags only when you already hold choreography worth recording verbatim - workarounds the CLI didn't surface (silently-dropped flags, undocumented params, pagination tricks, payload gotchas). Prefer the integrated one-call form - record the resource learning and the playbook in the same teach invocation:

bash
# Common case: record both the resource learning AND the playbook in one call.
forkable-pp-cli teach \
  --query "<user's question>" \
  --resource <id> \
  --playbook-file ~/playbooks/<shape>.json \
  --playbook-notes-file ~/playbooks/<shape>-notes.md
# (append shell `&` to background it)

# Alternate: playbook-only (no resource to record alongside).
forkable-pp-cli teach-playbook \
  --query "<user's question>" \
  --playbook-file ~/playbooks/<shape>.json \
  --notes-file ~/playbooks/<shape>-notes.md

Playbook files are JSON with steps, entity_slots, expected_tool_calls. Notes files are markdown carrying the gotchas verbatim. File-free callers (MCP-only agents) pass the same content inline: --playbook-json and --playbook-notes on the integrated teach form, --playbook-json and --notes on teach-playbook. On the integrated teach form, the playbook flags are optional - omit them entirely for a resource-only teach. On the standalone teach-playbook form, at least one of the playbook and notes flags must be set; both empty is rejected. Playbooks are keyed on the structural query family (entities stripped) so a recipe taught from one entity-shaped query applies to every other query of the same shape, with slots_resolved binding the live query's canonical at recall time.

When you DO find a playbook on a future recall, treat it as ground truth: replay the steps with slots_resolved substitutions, skip the discovery that the choreography already documents, and read notes before any step.

Step 6: playbook amend & when your debug response identifies a correction

If your debug-protocol response identifies a concrete correction the notes or playbook should know — a workaround, an undocumented endpoint shape, a stale field name, observed schema drift, an empty-payload fallback — fire playbook amend BEFORE emitting your user-facing response. Same fire-and-forget posture as teach.

bash
forkable-pp-cli playbook amend \
  --query "<exact recall query string>" \
  --add-note "<your concrete correction>"
# (append shell `&` to background it)

What counts as worth amending: a behavior you OBSERVED this session that future-you would benefit from knowing. Examples worth amending:

  • A workaround for a CLI surface that silently drops or misorders a flag.
  • An undocumented endpoint shape (response wrapped in {meta, results}, payload nested two levels deeper than the docs claim).
  • Observed schema drift (a field renamed, an index that shifted between seasons, a category label that the API now returns lower-cased).

What does NOT belong in notes:

  • The year-specific or entity-specific answer to the user's question. That's the response, not a learning.
  • Per-team / per-athlete / per-row data the playbook already retrieves at runtime.
  • Statements that paraphrase what the existing notes already say.

The amend command appends to the family's existing notes with a timestamped marker ([amend YYYY-MM-DDTHH:MMZ]: <text>). Multiple amends accumulate; the audit trail is visible. If no playbook exists yet for the family, amend creates a notes-only one (so cold-start corrections still land).

PII discipline for amend notes

playbook amend notes are designed to potentially flow upstream as shared knowledge in future versions of the Printing Press. Keep them clean of user-identifying content so the upstream-contribution path stays open without retroactive scrubbing:

  • Do NOT embed paths to user filesystems, personal API keys or tokens, user email addresses, user GitHub handles, or specific query histories tied to a single user.
  • Acceptable: endpoint shapes, undocumented field names, API gotchas, observed schema drift, workarounds for CLI surfaces, generalizable pagination or retry tactics.

If a correction is only meaningful with user-specific context, it belongs in a personal note, not in the playbook amend.

Measuring the loop

forkable-pp-cli learnings stats reports recall hit rate, teach-to-reuse, playbook resolution rate, and candidate confirm/reject counts from the local learn_events table. Rates are null until they have a denominator; everything stays on this machine. Use it to check whether the loop is earning its keep for this CLI.

Disabling learning
  • --no-learn on a single command short-circuits both recall and the teach write path. Use for deterministic agent flows or tests that must not be affected by accumulated learnings.
  • FORKABLE_NO_LEARN=true in the environment globally disables the pipeline.

Agent Feedback

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

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

Entries are stored locally as feedback.jsonl under the resolved data dir. They are never POSTed unless FORKABLE_FEEDBACK_ENDPOINT is set AND either --send is passed or FORKABLE_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.

This is CLI-tool feedback, not meal feedback. If a delivered meal itself had a problem (missing item, wrong item, quality), use meal feedback <deliveryId> --category ... --confirm instead (see Command Reference) — unlike this command, it actually reaches Forkable (a real GraphQL mutation for missing-item/missing-side, or a real contact-form POST for everything else), tied to a deliveryId/piece/orderId so the report is actionable. This feedback command stays local-only; meal feedback does not.

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:

SinkEffect
stdoutDefault; 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 or recurring agent reuses the same saved flags while providing different input each run.

forkable-pp-cli profile save briefing --json
forkable-pp-cli --profile briefing buffet-addresses --query example-value
forkable-pp-cli profile list --json
forkable-pp-cli profile show briefing
forkable-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

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

Argument Parsing

Parse $ARGUMENTS:

  1. Empty, help, or --help → show forkable-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

  1. Install the MCP server:
    bash
    go install github.com/mvanhorn/printing-press-library/library/food-and-dining/forkable/cmd/forkable-pp-mcp@latest
  2. Register with Claude Code:
    bash
    claude mcp add forkable-pp-mcp -- forkable-pp-mcp
  3. Verify: claude mcp list

Direct Use

  1. Check if installed: which forkable-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:
    bash
    forkable-pp-cli <command> [subcommand] [args] --agent
  4. If ambiguous, drill into subcommand help: forkable-pp-cli <command> --help.

© mvanhorn, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in cli-skills/pp-forkable of mvanhorn/printing-press-library.

Open the folder on GitHubat commit 7638ad4

Compare with similar skills

Pp Forkable next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Pp Forkable compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Pp Forkable this skillmvanhorn/printing-press-library2.1k—~9.7kAutomated safety check: NotesApache-2.0
MCP Server Builderanthropics/skills180k62 repos~2.3kAutomated safety check: PassApache-2.0
MCP Server BuildershareAI-lab/learn-claude-code78k5 repos~1.2kAutomated safety check: PassMIT
MCP Integration for Pluginsanthropics/claude-plugins-official37k11 repos~3.1kAutomated safety check: PassApache-2.0
Fastmcp Client CLIPrefectHQ/fastmcp28k1 repos~823Automated safety check: PassApache-2.0
MemPalace Memory SearchMemPalace/mempalace59k—~1.4kAutomated safety check: PassMIT

Similar skills

  • MCP Server Builder

    anthropics/skills

    Official

    Guides the design and implementation of Model Context Protocol servers in TypeScript or Python, from tool naming and error messages to evaluation.

    180k GitHub starsUsed in 62 repos~2.3k tokens
    Agent WorkflowsAuto-check passed
  • MCP Server Builder

    shareAI-lab/learn-claude-code

    Walks through building MCP servers in Python or TypeScript that expose tools, resources and prompts to Claude, with templates, registration and testing.

    78k GitHub starsUsed in 5 repos~1.2k tokens
    Agent WorkflowsAuto-check passed
  • MCP Integration for Plugins

    anthropics/claude-plugins-official

    Official

    Explains how to bundle Model Context Protocol servers in a Claude Code plugin, covering config files, stdio, SSE, HTTP and WebSocket server types, and authentication.

    37k GitHub starsUsed in 11 repos~3.1k tokens
    Agent WorkflowsAuto-check passed
  • Fastmcp Client CLI

    PrefectHQ/fastmcp

    Query and invoke tools on MCP servers using fastmcp list and fastmcp call.

    28k GitHub starsUsed in 1 repo~823 tokens
    Agent WorkflowsAuto-check passed
  • MemPalace Memory Search

    MemPalace/mempalace

    Mines project files and conversation exports into a local, searchable memory palace and recalls past work by semantic search through the mempalace CLI.

    59k GitHub stars~1.4k tokensUpdated yesterday
    Agent WorkflowsAuto-check passed
  • Crush Configuration

    charmbracelet/crush

    Explains how to configure the Crush coding agent with crushrc or crush.json, covering providers, models, LSPs, MCP servers, hooks, permissions and config precedence.

    29k GitHub stars~3.7k tokensUpdated today
    Agent WorkflowsAuto-check passed

More from mvanhorn/printing-press-library

All 505 skills in this repo
  • Agent Desktop

    mvanhorn/printing-press-library

    Desktop automation through the real Rust agent-desktop CLI, published in Printing Press through a small bridge.

    2.1k GitHub stars~2.3k tokensUpdated yesterday
    Auto-check: notes
  • Gfonts

    mvanhorn/printing-press-library

    Search, browse, and download Google Fonts from the terminal via the gfonts CLI.

    2.1k GitHub stars~574 tokensUpdated yesterday
    Auto-check passed
  • Pp 1688

    mvanhorn/printing-press-library

    The free, offline Trigger phrases: search 1688 for, find a factory on 1688 for, wholesale price on 1688 for, who is the cheapest supplier on 1688 for, compare 1688 suppliers for, use 1688, run 1688.

    2.1k GitHub stars~3k tokensUpdated yesterday
    Auto-check: notes
  • Pp Activity Japan

    mvanhorn/printing-press-library

    Inspect known Activity Japan plan IDs or URLs, compare dated prices and sessions, check language-sitemap coverage, and hand off to canonical booking pages.

    2.1k GitHub stars~2k tokensUpdated yesterday
    Auto-check: notes
  • Pp Adminbyrequest

    mvanhorn/printing-press-library

    Every Admin By Request portal action, plus a local SQLite mirror of audit, events, inventory and requests for ad-hoc...

    2.1k GitHub stars~3.3k tokensUpdated yesterday
    Auto-check: notes
  • Pp Agent Capture

    mvanhorn/printing-press-library

    macOS screen capture, window recording, GIF conversion, and agent evidence bundles from the terminal.

    2.1k GitHub stars~1.6k tokensUpdated yesterday
    Auto-check: notes

Categories

Questions about Pp Forkable

What does Pp Forkable do?

A CLI and MCP server for your Forkable office-lunch program, with a local database and history, spend, and preference queries the web app cannot answer, plus commands to set, confirm, and skip meal…. Pp Forkable is an agent skill from mvanhorn/printing-press-library. A CLI and MCP server for your Forkable office-lunch program, with a local database and history, spend, and preference queries the web app cannot answer, plus commands to set, confirm, and skip meal orders (dry-run by default; --confirm to apply).

When should I use Pp Forkable?

Pp Forkable fits situations like: phrases: what have I eaten on forkable; forkable lunch spend this month; which forkable venues do we use most; did my forkable meals match my dietary preferences.

How do I install Pp Forkable in Claude Code?

Run `npx skills add mvanhorn/printing-press-library --skill pp-forkable -a claude-code`. Or copy the skill folder (cli-skills/pp-forkable in mvanhorn/printing-press-library) into .claude/skills/pp-forkable in your project. Claude Code loads it when a task matches its description.

How do I install Pp Forkable in Codex?

Run `npx skills add mvanhorn/printing-press-library --skill pp-forkable -a codex`. Or copy the skill folder (cli-skills/pp-forkable in mvanhorn/printing-press-library) into .agents/skills/pp-forkable in your project. Codex loads it when a task matches its description.

Can I use Pp Forkable in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add mvanhorn/printing-press-library --skill pp-forkable -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/pp-forkable, .gemini/skills/pp-forkable, .github/skills/pp-forkable and .opencode/skills/pp-forkable in your project.

What does Pp Forkable need to run?

Going by SKILL.md and its folder, Pp Forkable needs the command-line tools its instructions call (go, claude and npx). Our summary lists: Node.js. Its frontmatter pre-approves these tools: Read, Bash.

Does Pp Forkable access the network?

SKILL.md contains no URLs. Its commands use npx, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Pp Forkable safe to install?

Our automated static check of SKILL.md found notes only (pre-approves every shell command (allowed-tools: bash)), nothing it rates as a warning. It is not a guarantee. Review the folder before installing.

What licence does Pp Forkable use?

Pp Forkable is published under the Apache-2.0 licence (declared in SKILL.md). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Pp Forkable use?

About 9.7k tokens (SKILL.md is roughly 39k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Pp Forkable?

Skills that share tags, products or a category with Pp Forkable: MCP Server Builder (anthropics/skills, 180k stars), MCP Server Builder (shareAI-lab/learn-claude-code, 78k stars), MCP Integration for Plugins (anthropics/claude-plugins-official, 37k stars) and Fastmcp Client CLI (PrefectHQ/fastmcp, 28k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Pp Forkable?

mvanhorn (a GitHub user) maintains it in mvanhorn/printing-press-library, which has 2,053 GitHub stars. The repository holds 505 skills in this directory. The repository was last updated on October 6, 2026.

Source: mvanhorn/printing-press-library on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.