Find the open-source solution that fits.

palamedes combines multi-agent research prompts, an agent-loadable skill, calibration scenarios, a browser UI, and offline report templates around a strict evidence discipline. Load-bearing claims need retrievable source text and verbatim quotes, while missing evidence is marked instead of filled from intuition; agreement among agents does not count as independent support. Stakes determine the depth of the workflow, and the browser UI uses the user's own API key.

job-hunt-skill turns a job description and an experience bank into a one-page tailored CV, a cover letter, a short outreach message, application answers, a submission tracker, and feedback rules across eight phases. The master CV remains untouched, metrics must be confirmed, and the final Submit action stays with the applicant. It began with UK AI startup roles, so senior, non-technical, or other-market searches need adaptation.

past-paper-topic-analyzer reads a chosen window of exam PDFs end to end, clusters topics inductively, and exports a three-sheet Excel workbook with coverage, every question grouped by topic, and a year-by-topic frequency heatmap. It counts an essay once, records cross-topic material separately, and preserves enough question detail to defend the tally. The workflow depends on accessible public PDFs and an explicit board, level, section, and year scope.

skill-triage turns a task into a routing plan with a restatement, complexity tier, risk flags, ranked skill choices, rejected options, execution order, and a proceed-or-ask verdict. Destructive or irreversible work always gets a confirmation gate, while an empty local match can trigger scrubbed discovery from a small registry allow-list. The tool is advisory: it recommends a route but does not block the caller or run the task itself.

theaccessible converts a PDF through theaccessible.org API into an accessible HTML rendition, a tagged PDF, and a veraPDF PDF/UA-1 conformance report. It returns download links and explains failing rules such as missing tags or alternative text instead of claiming compliance prematurely. The skill requires an API key and outbound HTTPS.

claude-tabletop inspects a project's files to choose a relevant incident or continuity scenario, then creates a facilitator runbook and an interactive HTML version. The runbook supports a timer, staged inject reveals, decision logging, and JSON state export. A companion skill turns that export into an After-Action Report with a timeline, gaps, and severity-tagged actions. It covers nine domains, including security, SRE, privacy, data/ML, and AI safety.

sd-coach turns Claude Code into a system-design interview coach for L3–L4 software roles. Choose a teacher-led coach, timed mock, Socratic tutor, rapid drill, or concept teach mode; sessions cover scoping, capacity, high-level design, APIs, schemas, low-level design, and failure modes. Mock sessions produce rubric-based scorecards, while Teach writes reusable notes to a personal concepts wiki.

OpenOrder helps an AI agent maintain a local investment-research wiki instead of leaving insights in chat history. It reads an index before answering, ingests pasted articles or earnings material into immutable raw files, updates company and industry pages, logs every write, and can lint for contradictions or stale gaps. The same Markdown-based wiki works across Claude Code, Cursor, Codex, and other supported agents.

webEmbedding starts with a URL, checks whether the page can be reused directly, and otherwise captures browser state before generating a bounded HTML, CSS, React, or Next.js scaffold. It compares the original and candidate through screenshots, DOM, computed styles, interaction states, and responsive breakpoints. The tool is aimed at static or semi-static pages; login-only surfaces, captchas, payments, and server-side app behavior remain out of scope.

cursor-chats-bridge connects Claude Code to Cursor's local chat database in read-only mode on Windows. An MCP server exposes workspace and chat listing, active-chat retrieval, search, and journal queries, while a scheduled poller records new messages every five minutes for later recap. It never writes to Cursor's database, but it only sees live messages from the currently open chat and does not backfill older history.

army-to-civilian helps service members translate U.S. Army experience for civilian resumes, LinkedIn profiles, cover letters, and interview answers. It first adapts the framing to the target audience and document type, then checks acronyms, tone, and common veteran self-translation mistakes before producing the draft. Its rank and MOS mappings are Army-specific, so other branches need adaptation.

The SendFlow API Skill teaches Claude Code to work with SendAPI's 44 WhatsApp-automation endpoints across campaigns, groups, messages, accounts, templates, blocking, verification, and media. It includes an OpenAPI 3 specification, real response examples, and Node.js and Python helper clients with authentication and rate-limit handling. The service is aggressively rate-limited and requires a user-created API key.

claude-caffeinate adds a cross-platform /caffeinate command for keeping the system awake while a long Claude Code session runs. It supports system-only or system-and-display modes, optional time limits, stop, and status commands. The implementation uses native Windows, macOS, or Linux mechanisms, stores one temporary state file, and scopes the wake lock to a process so it clears when that process exits.

feishu-service-bot-skill is a decision guide for self-hosted Node.js Feishu or Lark IM bots built on WebSocket. It addresses mention filtering, environment-based configuration, multi-provider LLM switching, in-process deduplication, a single-worker queue, image handling, lightweight sessions, hot-loaded subskills, and error reporting back to chat. It is aimed at internal bots for roughly 1–50 users, not high-concurrency or public webhook deployments.

cic-skill lets Claude analyze a concrete situation through six mindset models distilled from Guannaxi CIC's public story: craft as a moat, self-earned power, overseas leverage, moving on from a peak, persistence, and contrast through sincerity. It is designed for independent-business and career questions, not as a literal voice imitation, and the project frames the material as educational rather than official statements.

xhs-cdp-scraping connects Claude Code to a user's already logged-in Chrome through CDP to capture Xiaohongshu posts and comments. It relies on the page's own request signing, includes guidance for rate-limit and soft-block signals, and covers comment pagination and pseudonymized storage. The intended use is small-scale personal research with a burner account; high-volume scraping, automated login, captcha handling, and production commercial use are out of scope.

cc-passenger-prototype is a project-specific Claude skill for extending the CC ride-hailing passenger prototype. It can add new business pages, create variants, introduce BEM blocks, or maintain the token and base CSS files while preserving the established OKLCH, BEM, and WebView H5 language. It expects the project's source styles to exist and rejects Tailwind rewrites, third-party app imitation, and direct Vue, React, or UniApp output.

Use this Claude Code skill when several ideas, plans, or vendors need a second opinion. It gives each option an independent advocate, then asks a fresh judge to score the matchup against a rubric and verify interpretive claims with primary sources. Prompts, responses, searches, and grades are saved for an auditable single-elimination verdict; it is not designed to produce a full ranking.