菜单
Update an existing Can Agents Use catalog record by pull request. Use when a user wants to fix tool info, refresh launch signals, update API/CLI/MCP/browser support, revise pricing/auth/docs/sandbox evidence, or fix a logo for a tool already listed in phuctm97/canagentsuse.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Submit a new tool to Can Agents Use. Use when a user wants to add, suggest, contribute, or open a pull request for a missing agent-friendly API, CLI, MCP server, browser-operable product, docs source, pricing page, sandbox, or tool record in phuctm97/canagentsuse.
Use Can Agents Use, also searchable as canagentsuse, when any AI agent needs a skill to find agent-friendly tools, APIs, CLIs, MCP servers, browser workflows, sandboxes, pricing, docs, setup steps, and safety guardrails.
Compare agent-friendly tools with Can Agents Use. Use when a user asks for the best tool, ranked alternatives, side-by-side tradeoffs, evidence-backed recommendations, or which API, CLI, MCP, browser, pricing, docs, sandbox, and account setup profile is safest for an AI agent.
Find agent-friendly tools with Can Agents Use when a user asks which software, SaaS, API, CLI, MCP server, or browser workflow an AI agent can actually operate safely. Use for general tool discovery, alternatives, shortlists, and evidence-backed recommendations.
Find API-first and API-capable tools with Can Agents Use. Use when a user asks for software an agent can call through HTTP APIs, SDKs, webhooks, OpenAPI specs, service accounts, API keys, sandboxes, or machine-readable docs.
Find browser-operable tools with Can Agents Use. Use when a user asks whether an agent can use a web app, dashboard, admin console, signup flow, pricing page, documentation site, or browser automation workflow safely.
| name | update-agent-friendly-tool |
| description | Update an existing Can Agents Use catalog record by pull request. Use when a user wants to fix tool info, refresh launch signals, update API/CLI/MCP/browser support, revise pricing/auth/docs/sandbox evidence, or fix a logo for a tool already listed in phuctm97/canagentsuse. |
| allowed-tools | WebFetch |
Use this skill when the user wants to update a tool that is already in the Can Agents Use catalog.
For missing tools, use submit-agent-friendly-tool instead.
Use one of these paths:
https://canagentsuse.com/tools/<slug>, then choose Update existing tool.https://canagentsuse.com/tools/<slug>, then choose Copy update prompt.phuctm97/canagentsuse, update one existing source file in data/tools/<first-letter>/<tool-slug>.json, optionally add or fix one custom SVG logo at packages/website/public/logos/tools/<tool-slug>.svg, and open a PR with .github/PULL_REQUEST_TEMPLATE/update-tool.md.catalog/update-<tool-slug>.data/README.md, data/catalog.schema.json, data/taxonomy.json, and nearby records in data/tools/.data/tools/<first-letter>/<tool-slug>.json. Use the existing slug and path unless the PR is explicitly correcting the slug.submit-agent-friendly-tool.launchSignals, use exact public numbers when available, null for unavailable numeric signals, unknown only where allowed, and conservative tiers instead of fake or inflated values.packages/website/public/logos/tools/<tool-slug>.svg and set logoPath to /logos/tools/<tool-slug>.svg only when the SVG exists..changeset/*.md version plans when this should release the website or CLI. Do not edit package versions, lockfiles, changelogs, release workflows, repo scripts, app code, or agent skills manually.main.In the PR body, explain:
Use update mode for:
launchSignals.Run:
bun run catalog:build
bun run catalog:format
bun run catalog:audit
bun run build
git diff --check
data/catalog.json, add secrets, database credentials, Docker/Neon setup, generated build output, or unrelated refactors..changeset/*.md files when the change should release the website or CLI.null, unknown, conservative tiers, or clear caution notes when evidence is missing.