kendreaditya/interviewcoder

Interview Coder

Read-only CLI for interviewcoder.co. Wraps scripts/interviewcoder.py, a single-file Python tool that scrapes the React Server Components (Flight) payload embedded in each page. No login, no cookies, no encryption.

Pair with the blind skill for opinion/chatter on the same company — this skill is the structured-question source.

Invocation

The script is on $PATH as interviewcoder (symlinked from ~/.config/scripts/). Use that short form in examples. Falls back to python3 scripts/interviewcoder.py if invoking from a fresh environment.

Standard library only — no pip install step.

interviewcoder --help

Commands

list — page of questions (filterable)

interviewcoder list \
    [--company NAME] [--job-type TYPE] [--stage STAGE] \
    [--outcome OUT] [--sort SORT] [--page N] [--limit N] [--json]

Return one page of posts. Each post carries title, company, position, period, job type, interview stage, author, engagement counts, a content snippet, and the structured questions (with difficulty and tags).

Examples:

interviewcoder list --company Apple --limit 5
interviewcoder list --company Amazon --sort most-bookmarked
interviewcoder list --job-type Intern --limit 5
interviewcoder list --company ByteDance --page 2

Known filter values seen in the wild:

• --job-type: Fulltime, Intern
• --stage: Phone Screen, OA, Onsite, Technical Interview, HR
• --sort: oldest, most-bookmarked (default is newest)
• --company: use the value as it appears on the site (e.g. Apple, Amazon, ByteDance, Capital One, Capitalone, Applied-Intuition, Boston-Consulting-Group). Run companies first if unsure of the exact spelling — some companies appear under multiple aliases.

The output header shows page X/Y so paginate with --page.

read — fetch one post + its structured questions

interviewcoder read <id-or-url> [--json]

Accept a bare UUID or any URL containing one (the /questions/{uuid} path). Render the full post body (markdown), metadata, and each structured question with difficulty, tags, and leetcode_ref.

interviewcoder read c2e84e3e-a0ee-415e-ae9f-db56988b85fb
interviewcoder read https://www.interviewcoder.co/questions/c2e84e3e-a0ee-415e-ae9f-db56988b85fb

If hasAccess is false (the site paywalls some features for logged-out users) the script prints a note. The post body is still returned in full — only UI features like bookmarking are gated.

companies — company directory with counts

interviewcoder companies [--limit N] [--json]

List every company with at least one post, ordered by count descending. Run this before list --company if uncertain of the exact spelling.

search — client-side search across recent pages

interviewcoder search "<query>" [--company NAME] [--pages N] [--limit N] [--json]

No server-side full-text search exists, so this fetches the first --pages pages (default 3) and substring-matches against title, content snippet, question titles, and tags. Combine with --company to narrow before searching.

interviewcoder search "system design" --company Amazon --pages 2
interviewcoder search "binary search" --pages 5 --limit 10

Slow for wide scans — each page is one HTTP fetch.

export — dump one post to md or json

interviewcoder export <id-or-url> [--format md|json]

Write a self-contained Markdown (default) or JSON document to stdout.

interviewcoder export c2e84e3e-... --format md > apple-phone.md
interviewcoder export c2e84e3e-... --format json > apple-phone.json

Common flags (work on every command except export)

• --json — structured JSON instead of pretty text. The post schema is the Post dataclass (see below). Pipe into jq.
• --limit N — cap output to N items.
• -v / --verbose — log fetched URLs to stderr.

Common flags must appear after the subcommand name, e.g. interviewcoder list --company Apple --limit 5.

How to choose the right command

| User intent | Command | |---|---| | "What questions have been asked at {company}?" | list --company {Name} | | "Show me {company}'s top bookmarked writeups" | list --company {Name} --sort most-bookmarked | | "Most recent intern interviews" | list --job-type Intern | | "Show me the full writeup at {url}" | read {url} | | "Which companies are covered? How many posts each?" | companies | | "Find writeups mentioning system design / binary search / etc." | search "{query}" | | "Save this writeup for later" | export {url} --format md > file.md |

Data schema

Post (returned by list, search, and as .post from read): id, title, author, company, position, period, job_type, user_type, interview_stage, outcome, content, posted_at_original, created_at, url (canonical interviewcoder.co URL), source_url (1point3acres etc.), category, views_count, likes_count, bookmarks_count, comments_count, questions[].

Question (inside Post.questions): id, question_number, title, description, difficulty (Easy / Medium / Hard / null), leetcode_ref, tags[], created_at.

Article (returned by read): post (Post), markdown (the formatted long-form body), has_access (bool), is_authenticated (bool).

Limitations

• First page only on list per call — use --page N to paginate.
• No server-side search — search fans out across pages client-side. Slow for wide scans.
• No write operations. No login, no posting, no bookmarking. Logged-in features (hasAccess=true posts, personal bookmarks) would require defeating the auth flow — out of scope.
• Posts are scraped from 1point3acres. The source_url field on each post points at the original thread; the interview text is often translated and reformatted.

Error handling

The script exits with:

• 0 on success
• 1 on errors (network, parse failure, etc.)
• 2 on "not found" — bad post id or 404