2 Commits

Author	SHA1	Message	Date
replicas-connector[bot]	010ae38779	Playbooks for parallel and exa, and a 20x price error in the stage map (#107) ... * Playbooks for parallel and exa, and a 20x price error in the stage map Section 11 claims every credits-based provider in the catalog has a playbook. Nine do not: brightData, builtwith, datachimp, exa, openRouter, parallel, proxycurl, sillage and x. This adds the two that matter most for research and leaves the claim to be squared with the other seven separately. Both are worth having because both beat what the recipes currently reach for. parallel.extract reads a known URL for 0.025, half of firecrawl.scrape. parallel.createTask is the only action in the catalog that fills a caller-supplied output schema, and at processor lite it is 0.125 against linkup.instruct at 1. exa.search is the only rung with a document-type category filter and publication-date bounds, which is what makes it worth 0.175 fixed plus 0.025 an item over a 0.05 flat search. Every action, input and price here was read from the live integration catalog rather than from documentation, per the lesson this pack keeps relearning: a capability claim needs the implementation, not the docs index. That is also how the error surfaced. references/stage-action-map.md priced serper.search at 1. It is 0.05, and provider-playbooks/serper.md had it right throughout, so the map was steering agents away from the cheapest search rung by a factor of twenty. alternatives.md carried the same wrong number for serper.searchPlaces. Both corrected, and the correction is stated in the table rather than made silently. The Web research table is rebuilt around the actual decision: known URL means parallel.extract, plain keyword means serper, document-type or date filter means exa, structured output means parallel.createTask at lite. The createTask ladder gets its own table and a warning. processor is required with no default, and the top rung is 60 credits per record, which is more than a full contact waterfall. Co-Authored-By: Auree Aubert <16732193+aureeaubert@users.noreply.github.com> Co-Authored-By: replicas-connector[bot] <replicas-connector[bot]@users.noreply.github.com> * Ban datachimp in code, and close three of the playbook gaps Two things, both about the same claim: section 11 said every credits-based provider has a playbook, and nine did not. Datachimp is removed everywhere and cannot come back. Aurelien's instruction, 2026-08-15. It appeared in the generated catalog and in two lines of cargo-storage prose. Deleting those would have lasted until the next --refresh, so the removal is enforced in code instead: a DENIED_INTEGRATIONS set filters it out of the catalog on both write and read, and skills-lint fails on the string anywhere in the repo, which is what catches the hand-written half no generator touches. It was also an integration with zero callable actions, so nothing was lost. Three playbooks added, each earning its place on price rather than on coverage. builtwith's getDomainSummary is FREE and answers most stack questions, so the playbook's whole argument is that it runs in front of enrichDomain (1); the current recipes reach straight for a paid rung. sillage's searchLeads is also free and is the only free signal rung in the catalog, which makes it first on any signal question by definition. x is fourteen actions at 0.02, the cheapest signal surface there is, and the playbook spends most of its length on why that is dangerous rather than on how to use it. Three are deliberately not written and the claim now names them instead of overstating. openRouter exposes a model lister and no credits-based actions, so there is nothing to document. proxycurl is excluded on instruction. brightData is excluded because its entire action set is consumer social-platform scraping (Instagram, TikTok, Facebook), which acceptable-use already refuses for person targeting: documenting it as a GTM provider would contradict the gate two files away. Every action, input and price read from the live integration catalog. Co-Authored-By: Auree Aubert <16732193+aureeaubert@users.noreply.github.com> Co-Authored-By: replicas-connector[bot] <replicas-connector[bot]@users.noreply.github.com> * x: a cost table a validator can read, not a grouping a human can The action list was grouped by input shape, which reads well and parses to nothing. Every other playbook here carries `| action | cost | inputs | use for |` and the downstream validator in getcargohq/gtm-skills asserts a skill's quoted price against that column, so a skill calling any x action would have failed with "not in that provider's cost table" and sent someone hunting for a missing price rather than a missing table shape. The four engagement-list actions keep their warning in the Use for column instead of only in the anti-patterns, since that column is what a reader scans. Co-Authored-By: Auree Aubert <16732193+aureeaubert@users.noreply.github.com> Co-Authored-By: replicas-connector[bot] <replicas-connector[bot]@users.noreply.github.com> * Install the whole bundle by default, and stop hand-counting installs skills.sh counts per skill, and one `skills add` sends a single event carrying every skill it installed. So the command in the README decides how many counters a real install moves. `--all` is now the documented form: it is shorthand for `--skill '*' --agent '*' -y` and behaves the same in a terminal, in CI, and inside an agent, where a bare add would drop into an interactive picker. Two stale hand-maintained facts removed rather than corrected, which is the same call this repo made about "sixteen skills" in the CLI help. The install total read "installed 54,885 times". The live figure is 58,245 across the seventeen skills. A number in prose that only moves when somebody remembers to move it will be wrong again next week, and this one is a public claim about traction, which is the worst kind to be casually wrong about. The gtm-skills section listed twelve skills by name. That repo has grown past twelve, so the list now points at its README instead: a hand-maintained copy of someone else's list is a copy that goes stale. Co-Authored-By: Auree Aubert <16732193+aureeaubert@users.noreply.github.com> Co-Authored-By: replicas-connector[bot] <replicas-connector[bot]@users.noreply.github.com> --------- Co-authored-by: replicas-connector[bot] <replicas-connector[bot]@users.noreply.github.com> Co-authored-by: Auree Aubert <16732193+aureeaubert@users.noreply.github.com>	2026-08-14 23:58:36 -07:00
Auree Aubert	3708f8c955	Agent discoverability: trigger-shaped descriptions, routing evals, standalone bootstrap (#77) ... * Agent discoverability: trigger-shaped descriptions, routing evals, standalone bootstrap The `description` field is the only text an agent weighs before deciding whether to load a skill. Everything else in the bundle is invisible until that decision is already made — and twelve of the seventeen descriptions read as "Manage <CLI nouns> using the Cargo CLI", which matches a user who is already thinking in Cargo's object model rather than one stating a job. This rebuilds that layer and, for the first time, tests it. Routing - Every description rewritten to a four-part template: job in the user's words -> literal quoted trigger phrases -> proper nouns -> an explicit `Skip when:` pointing at the sibling skill that owns the adjacent case. - Integration and provider names are generated, not hand-maintained. sync-trigger-slugs.ts fills cargo-gtm's `Providers:` list from the playbook tree and cargo-connection's `Integrations:` list from a committed catalog snapshot (138 integrations, 50 credits-based). `--refresh` re-pulls from a logged-in CLI; `--check` gates CI. - evals/routing.jsonl (92 prompts) + routing-eval.ts, wired into CI. Structural tier asserts the template and that each skill keeps >=5 terms unique to it; lexical tier requires the expected skill to rank first; an opt-in --llm tier asks a real model to route from descriptions alone. Deep paraphrases are tagged "hard" and reported rather than gated — a lexical proxy cannot judge them fairly. Standalone installs - `skills add --skill <one>` is a supported path and the registry lists each skill separately, but thirteen skills' only install instructions were a relative link to ../cargo/references/prerequisites.md, which does not exist in a single-skill install. Every SKILL.md now has a self-contained Bootstrap section, enforced by the linter. Coverage - New capability skill cargo-segmentation — the segmentation CLI domain, previously uncovered while segments were load-bearing in most recipes. Verified against the live CLI, including the flag traps. - Three cargo-gtm recipes: source-planning (probe and cost candidate sources per *hit* before any fan-out), ads-audience-activation (Google Customer Match / LinkedIn Matched Audiences), review-and-iterate (human review of judgment output, kept as an eval set). - Router gains a "these skills vs a workspace MCP server" section: Cargo has no first-party MCP server — a workspace builds one and serves it over `cargo-ai mcp` — so the routing rule is scale and reproducibility. Index and hygiene - llms.txt now lists every recipe and provider playbook with a generated summary: 17 entries -> ~80 job-named addressable URLs. - Router trimmed ~80 lines of per-skill summaries that restated each skill's own description; critical rules kept in full. - Linter: bootstrap check, skill-count claims validated against the tree, `doctor` and `mcp` added to the known CLI domains. - CONTRIBUTING.md, keyword-dense README opening, updated repo metadata. Also carries the aiArk/apolloio priority-stack promotion authored in a parallel session on this branch (cargo-gtm 1.12.0) — the two touch the same SKILL.md files and could not be split cleanly. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> * Surface the 100-credit free tier where an agent meets it There is no purchase gate between installing Cargo and producing a real deliverable — a new account starts with 100 free credits and no card — and nothing in the bundle said so. An agent that doesn't know the budget can't tell a first-time user whether the demo is a look around or a commitment, and defaults to hedging. Now stated at every point an agent encounters the question: - README opening and the generated llms.txt install block: an agent can install, sign the user up, and hand back a list in one turn. - Router install section and prerequisites.md, with anchors for what 100 credits buys (~5,000 sourced leads, ~1,000 verified-email enrichments, ~50 fully enriched contacts). - cargo-quickstart: the demo spends ~0.5 of the 100, and the skill now says so before the first paid call — that reframes the moment from a purchase decision into a look around, which is the job of a quickstart. - cargo-billing: a "what 100 credits buys" table, plus two new trigger phrases so free-tier questions route here at all. - cost-discipline.md: on a new account, receipts and next-step proposals must be framed against the free tier. "87.6 of your 100 free credits left" is a number the user can act on; "87.6 remaining" is not. Also warns that phone lookups (6-7/call) are the fastest way to burn a tier a first session should end with something left to try. Two routing eval cases added for free-tier questions (79/79 core). Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> * Run the routing evals against a real model in CI The lexical tier passes 79/79, but it is a deterministic proxy written alongside the descriptions it grades — it cannot fail in a way that surprises us, which is what makes it weak evidence. The only honest scoreboard for the description rewrite is whether a real model, shown nothing but the 17 descriptions, routes the eval prompts correctly. That tier existed but had never run. Now it does, in its own workflow. - New .github/workflows/routing-evals-llm.yml, using the repo's ANTHROPIC_API_KEY. Deliberately not part of skills-lint: it costs money and is non-deterministic, so it must never stand between a typo fix and a merge. Triggers on PRs and pushes that touch a SKILL.md, the eval cases, or the runner; weekly on a schedule so model drift surfaces even when nobody edits a description; and on demand with a model override. - runLlm reworked for CI: 6-way concurrency (92 sequential calls is a slow job), retry with backoff on 429/5xx, tolerant answer parsing, and a core/hard/overall breakdown instead of one number. - Results are written to GITHUB_STEP_SUMMARY as a table, including every misroute with what the model picked instead — so a regression is readable on the run page without opening logs. - Optional --llm-min=<percent> floor, wired to the exit code. Left unset until the first runs establish a baseline; a floor guessed before the data would either never fire or block on noise. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> * Act on the first real-model routing run: 92/94, two fixes Baseline from the model tier (claude-haiku-4-5, judging from the 17 descriptions alone): 92/94 overall — core 79/79, hard 13/15. The lexical proxy's 3/15 on the hard tier was noise about the proxy, not the descriptions, which is what that tier existed to establish. Both misroutes were worth having: - "Our outbound is going to the wrong people — help me figure out who actually buys from us" routed to cargo-context instead of cargo-gtm. A real collision: cargo-context's triggers include "who do we sell to", which reads as ICP ownership. It owns *writing the conclusion down*; deriving it from won/lost data is cargo-gtm. Added that boundary to cargo-context's Skip when, and the two phrasings to cargo-gtm. - "Give the assistant access to our sales collateral so it stops making things up" routed to cargo-content instead of cargo-ai — and the model was right. As written the first step is getting the collateral into the workspace. Corrected the expectation and added the other side of the boundary as its own case ("attach the library we already uploaded to the agent"), which the lexical tier then failed, correctly: cargo-ai never said "library" or "release resources". Added both, so the content/ai split is now legible to a ranker and a model. Set --llm-min=93 now that a baseline exists. Deliberately below the observed 98%, not at it: a model tier that fails on one non-deterministic flip trains people to ignore it. Lexical 80/80 core. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> * Document cargo-ai mcp — the bridge that was written down nowhere Cargo reaches coding agents through two doors. Skills reach agents that implement skills.sh; a workspace MCP server reaches everything else — Claude Desktop, ChatGPT, and whatever ships next. The second door was fully built and documented in exactly one place: `cargo-ai mcp --help`. Worse, cargo-ai/SKILL.md taught `ai mcp-server create` (building a server) without ever mentioning the command that serves it, so an agent could construct one and have no idea how to connect it — and the section conflated publishing a server with consuming someone else's, which are unrelated surfaces. - MCP section rewritten around the two directions, with a table: publish (`ai mcp-server`, called by any MCP client, wired via `cargo-ai mcp`) vs consume (`ai mcp-client`, called by your agents, wired via `release update-draft --mcp-clients`). States plainly that there is no first-party Cargo MCP server — a workspace builds its own. - Documents `cargo-ai mcp` with the registration one-liner, the CARGO_MCP_SERVER_UUID fallback, and the stdout-is-protocol constraint. - Details verified against the live API rather than --help: actions take kind "tool" *or* "agent" (an agent is exposable as a callable MCP tool), resources take model or file, and `update` replaces the arrays wholesale instead of merging — a real footgun. - Routing rule for which door to use, pointing at the router section. - Two trigger phrases ("expose our tools as an MCP server", "use Cargo from Claude Desktop or ChatGPT") and two eval cases, so the publish direction is reachable from a description at all. It previously wasn't. Lexical 81/81 core. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>	2026-08-12 10:18:32 -07:00