Verify & Repair Algebra

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: none for the install itself: no account, no API key, no paid service, and nothing else to install first. The skill is workflow guidance only (three text files). To use it afterwards you need a ResiMind checker within reach — an available ResiMind MCP connection or an installed resimind command — and that runtime is NOT installed during this install: setting it up is a separate step, only if I later ask for it (its core runs on Python 3.10+; MCP needs Python 3.11+ and the optional SDK; the interface reference installs it from the versioned GitHub release source, not from PyPI). Scope: bounded rational polynomial algebra only — not calculus, not arbitrary theorem proving, not engineering certification. Install "verify-and-repair-algebra" for me. It is a verify-and-repair workflow for polynomial algebra: check a proposed identity or each step of a derivation with exact coefficient evidence from ResiMind, report the first rejected step, propose a correction from the mismatch, and recheck the repair against the original — at most three repair attempts. Repository: https://github.com/1105216375-alt/resimind (the skill is the whole skills/verify-and-repair-algebra/ folder at the pinned commit: SKILL.md, agents/openai.yaml, references/interface.md) Pinned version: install the code at commit 4055d9c8a065b687c4b13b04f3094d6c14545db4 exactly. Do not substitute a newer version. If this commit is unreachable, stop and tell me. 1. Clone https://github.com/1105216375-alt/resimind into a temporary folder, check out commit 4055d9c8a065b687c4b13b04f3094d6c14545db4, and copy the skills/verify-and-repair-algebra/ 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 55e9fcd39aff8f7626a322579ea7525a7f8921dda445ce2897e9409001bf1bd3 - agents/openai.yaml: SHA-256 a0bb2574670c3081af929648c0d49e129a0284277b7115047a12f9a931fd8be9 - references/interface.md: SHA-256 5fe19aca56788be3094f39f3877e97251d2e89d8a416b916d1f8466fb942bf44 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 every other file in the folder) 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 3 files only: SKILL.md (2,481 bytes), references/interface.md (1,658 bytes) and agents/openai.yaml (266 bytes). There is no code, no script and no dependency list in the folder: SKILL.md is the workflow, interface.md is the setup and request-syntax reference, and openai.yaml is display metadata (name, short description, default prompt) for OpenAI-style agents. - Nothing is sent anywhere during installation itself; the only download at install time is the skill folder from GitHub at the pinned commit. The files name no account, no API key, no telemetry and no remote service for the install. - At use time, the workflow preserves the original expression, declared variables and assumptions, and checks the work independently of the proposing model: it uses my chosen tools when I specify them, otherwise an available ResiMind MCP connection or its JSON-lines CLI. Both ResiMind routes are local (MCP over stdio, or one JSON object per line on stdin) and run offline with no model API key. SKILL.md states the workflow does not require publishing user inputs, modifying project configuration, or sending data to an external model. - The accepted grammar is narrow: integer literals, +, -, *, nonnegative integer powers and division by a nonzero integer literal. Conditions are never silently discarded, and a variable denominator is never turned into an unconditional polynomial identity. For a multi-step derivation, adjacent steps are checked in order and the first rejected step is reported; if a preceding step is unresolved, that uncertainty is reported instead of certifying the prefix. Outcomes are verified (report the exact coefficient evidence), rejected (use the mismatch to propose a correction; expand_polynomial can provide independently checked sparse terms), or unresolved (report the unsupported operation or work limit; this is not a disproof). A proposed repair is rechecked against the original source, at most three repair attempts, and an exact algebra check is never described as a Lean proof. Sparse coefficients are decimal strings: keep their exact values. - learn_polynomial is used only when the task includes retaining or learning rules: a verified solution alone does not establish that a new rule was learned (inspect learning.new_rule_ids), one server process stays alive for learning and recall because a restart clears this interface's process-local memory, and a recall miss permits another solver but is not reported as successful rule reuse. - Do not install the ResiMind runtime during this install. The interface reference carries a setup line for it (python -m pip install of the versioned release tarball https://github.com/1105216375-alt/resimind/archive/refs/tags/v0.16.0.tar.gz into an appropriate isolated environment — explicitly not an assumed PyPI package), plus a link to the full connection instructions on GitHub. That download and install happen only later, and only if setting up ResiMind is part of a task I actually give you. 3. When I say "yes, install", install it as a skill: copy the whole folder into the agent's skills directory, in a folder named "verify-and-repair-algebra", keeping every file (SKILL.md, agents/openai.yaml, references/interface.md). The upstream routes, npx [email protected] add 1105216375-alt/resimind --skill verify-and-repair-algebra or a manual clone followed by copying the skills/verify-and-repair-algebra folder into ~/.claude/skills/, install the same three files; either route is fine as long as the files match the fingerprints. 4. Verify with no network calls: frontmatter valid, files in place. 5. Report what was installed, where, and what I still need to do myself (for example: provide an available ResiMind MCP connection or an installed resimind command myself if I want to use the skill — setting that runtime up is a separate step with the download disclosed above, and it only happens if I ask for it; using the skill is a separate step that only starts when I ask for it). 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.