Compare Shows

Before you start: read this whole install prompt, tell me in a few lines what this skill does and what installing it involves, and wait for my go-ahead before you do anything else. Prerequisites — read this first, because this skill does nothing on its own: installing it only copies one instruction file (SKILL.md, no code, no scripts). To use it afterwards you must separately connect the remote Arcmira MCP server (https://mcp.arcmira.com/mcp) in your agent host and sign in to an Arcmira account through the host's OAuth flow (the host opens arcmira.com for sign-in and consent; a client that cannot do OAuth can instead send an arc_sk_ account key from arcmira.com as a Bearer key — never put that key in the skill folder). Arcmira reads are metered: they use credits from your account's plan first, then the on-demand budget you set in the Arcmira dashboard, and a read the plan or budget blocks fails with an error such as paid_plan_required or quota_exceeded. This install does not connect the MCP server, does not sign you in, does not create an account and does not buy anything: that connection and account are your own step, done by you, outside this install. Install "compare-shows" for me. It compares two podcasts or YouTube shows side by side over a stated window (the worked program in SKILL.md uses the last 90 days): it resolves each show name you give to its YouTube channel id, sizes each show (videos indexed and the date the index runs through), takes each show's latest episode, ranks what each show covers (organizations and products), reads what both shows covered from the overlap, and joins each show's sponsors on entity id to list the shared sponsors with ad reads per show. Every program runs through the Arcmira MCP tool arcmira_execute_read, with arcmira_describe as the method reference. A name that resolves only as a suggestion is an assumption and the answer must name it and quote the suggestion's evidence; when several entities fit, the program returns the options as a question to you and runs nothing else. Repository: https://github.com/arcmira/mcp (the skill is the whole plugins/arcmira/skills/compare-shows/ folder at the pinned commit: SKILL.md only, 1 file) Pinned version: install the code at commit 4598d894ad1a1856cbc10473b374d2f3aa7bc6c6 exactly. Do not substitute a newer version. If this commit is unreachable, stop and tell me. 1. Clone https://github.com/arcmira/mcp into a temporary folder, check out commit 4598d894ad1a1856cbc10473b374d2f3aa7bc6c6, and copy the plugins/arcmira/skills/compare-shows/ folder from that checkout. Then compare every file in the folder with the fingerprints recorded for this listing on 2026-10-19: [FINGERPRINTS] - SKILL.md: SHA-256 40f905f278ba803a70600fa1ff25a200033febd0a461e84e7868733657935359 If a file does not match, it changed since the listing was prepared: do not install it yet, tell me what you found and let me decide. 2. Read SKILL.md in full, and tell me in plain words: what the skill does when used, which services it connects to, whether it reads or sends any personal information or keys, and anything that seems unusual. Stop on any red flag and tell me before going further. Confirm these points from the files: - The skill folder holds 1 file only: SKILL.md (5,188 bytes). There is no script, no code and no other file in the folder, and nothing in it runs locally. Functionally it is NOT self-contained: it only teaches the agent to call the remote Arcmira MCP tools (arcmira_execute_read, and arcmira_describe for the method reference) — the MCP connection, the account and the plugin files that normally bundle that connection (plugins/arcmira/.mcp.json / mcp.json, the other five Arcmira skills) are NOT in this folder and are not installed by this install. - Nothing is sent anywhere during installation itself; the only download at install time is the skill folder from GitHub at the pinned commit. No program is run, no MCP tool is called, no key is asked for, no account is created and there is no telemetry in the folder. - What leaves this machine at use time, disclosed here before it ever happens — using the skill is a separate step that only starts when I ask you to compare two shows: (a) the agent sends ONE JavaScript program per question, as the code argument of arcmira_execute_read, to the remote Arcmira MCP server at https://mcp.arcmira.com/mcp, which runs it against https://api.arcmira.com on my Arcmira account, under the OAuth token (or arc_sk_ Bearer key) my host connected with — the program itself never holds the key. The worked program sends: the two show names I gave, my own words about each name as the resolve context (only words from my message, never a guess), the resolved YouTube channel (UC) ids of both shows, a date window built from arcmira.daysAgo(90) and arcmira.today() in the worked example, the entity types "organization" and "product", and a limit of 40 — through two resolve calls, two status calls, two episodes calls (latest episode, limit 1), one occurrences call carrying both channel ids, and two sponsors calls. (b) Arcmira logs every tool call to its product analytics (PostHog), linked to my account (this logging is documented in the repository README, "What we log"): one record per call holds the tool name, my host's name and version, the server version and how long the call took; the input — for an execute tool, the program text, first 4,000 characters — plus the optional intent (at most 300 characters: my request in a few words) when the agent sends it; and the outcome, not the result (ok or the error code, the result size, whether it was truncated, and the API routes made, not their query strings). Anything that looks like a credential or an email address is replaced with [redacted] before the record is stored; results, transcript text and program logs are never logged; the record is sent after the result. (c) Metering: the reads in the program use credits from my account's plan, then my on-demand budget; no tool takes a spending amount, and a blocked read returns paid_plan_required, quota_exceeded, spend_limit_exceeded, member_limit, fair_use_cap or on_demand_paused instead of data. (d) Feedback, at the end of a use: SKILL.md says to send ONE arcmira_feedback when anything was wrong, slow or missing for me, or when I want a source or capability Arcmira lacks — that tool sends a category (wrong_entity, bad_data, missing, slow, confusing or other), a note saying what happened, and optionally a request_id or call_id, to Arcmira (one POST /v1/feedback, stored as a record). Never send feedback during this installation, and at use time send it only for that stated case. (e) Monitors: this skill creates none and its program uses arcmira_execute_read only (no writes). SKILL.md's closing note only points out that a monitor can follow what a research turn found and that the separate company-watch skill sets one up; a monitor is a separate step I would have to ask for, and each monitor alert uses 25 credits from the account's plan, then its on-demand budget. - The local-only parts: there are none inside the skill folder beyond the instruction file itself — all research data comes back from Arcmira's index of YouTube and podcast transcripts. SKILL.md's answer rules stay between you and me: name both shows as resolved with their channel ids, give the side-by-side (videos indexed, indexed_through, latest episode, each show's top subjects with episode counts), the shared coverage from the occurrences "shared" list with episodes per show, and the shared sponsors with ad reads per show, for a stated window; never count episodes to size a show (status is the count), and an empty "shared" in a short window is an answer, not an error — widen the window only if I asked for a longer one. - Do not connect the Arcmira MCP server, do not sign in, do not create an Arcmira account, do not set or ask for an arc_sk_ key, do not run the worked program or any MCP tool, and do not send any arcmira_feedback during this install: connection, account and any comparison are my decision, later, when I actually ask for one. 3. When I say "yes, install", install it as a skill: copy the whole compare-shows folder into the agent's skills directory, in a folder named "compare-shows", keeping its 1 file (SKILL.md at its root). A manual clone followed by copying the plugins/arcmira/skills/compare-shows folder into ~/.claude/skills/ is the route described here; any other route is fine only as long as every installed file matches its fingerprint. 4. Verify with no network calls: frontmatter valid (name: compare-shows), all 1 file in place with matching fingerprint. Do not call arcmira_describe, arcmira_execute_read or any other MCP tool to verify. 5. Report what was installed, where, and what I still need to do myself (for example: connect the Arcmira MCP server at https://mcp.arcmira.com/mcp in my host myself and sign in to my own Arcmira account through OAuth — or configure my own arc_sk_ account key if my client cannot do OAuth — knowing the skill folder alone cannot work without that connection; make sure my plan and on-demand budget cover the metered reads; then, to use it, give the skill the names of two shows and a window if I want one other than the worked example's 90 days, and pick myself when a resolve returns several options; using the skill is a separate step that only starts when I ask for a comparison, and the use-time sends, logging and credit use disclosed above happen only then). GitHub is optional: if I have a GitHub account or the gh CLI, you may use it; otherwise public access is fine. Never require it unless it's in the prerequisites above. Rules: don't touch anything outside the temp folder and the install target. If anything looks off, stop and ask me.