Search verified, fresh upcoming/ongoing hackathons with optional filters.

date_range: upcoming | ongoing | upcoming_and_ongoing | all mode: online | offline | hybrid | any geography: free-text location filter (e.g. 'San Francisco', 'Worldwide') stack: comma-separated tech keywords to prefer prize_min: minimum total prize USD (0 to ignore) include_demo: explicitly include synthetic fixtures for local demos only

Discover public event-page URLs with TinyFish Search, then verify each one.

Search snippets are never presented as events. A candidate appears only after its page is fetched and contains hackathon evidence. Ingest a selected URL to store normalized metadata and make it eligible for later ranking.

Get full normalized metadata, including source and verification state.

Visit a public URL and require page evidence before trusting it as a hackathon.

This never infers validity from a URL shape. If hackathon_id names a non-demo stored record, its verification state is updated with this result.

List tracks and sponsor challenges for a hackathon.

Return rules, eligibility, judging criteria, and artifact requirements.

If refresh_from_web=True and FIRECRAWL_API_KEY is set, scrape the event page to enrich stored rules text.

Ingest a hackathon landing page via Firecrawl and store normalized metadata.

Requires FIRECRAWL_API_KEY. Without it, returns guidance and does not invent data.

Analyze a project from repo/demo/pitch/README and store a ProjectProfile.

Returns project_id for use with scoring and competitor tools.

Re-derive structured features for an existing stored project profile.

Score how well a project fits a hackathon with dimension breakdown + explanation.

Formula: 0.24theme + 0.18stack + 0.18feasibility + 0.16sponsor_fit + 0.14novelty + 0.10competitive_pressure_adjustment

Rank verified, fresh upcoming/ongoing hackathons for a project.

Set include_demo=True only for the bundled local walkthrough fixtures.

Recommend the best track/sponsor challenge and a positioning strategy.

Find whether a hackathon exposes a public project gallery and return its URL.

Competitor intelligence only uses intentionally public pages.

Extract publicly visible submissions for a hackathon gallery.

Uses TinyFish when TINYFISH_API_KEY is set and use_live_extraction=True. Cached data is available only with use_live_extraction=False and is always labelled cached. Never bypasses auth.

Cluster public submissions by idea similarity and track family.

Find nearest public submissions to the user project (similarity neighbors).

Estimate crowding / novelty risk for the project in a public gallery.

Suggest concrete differentiation moves based on public submission landscape.

Find past editions of this hackathon series and/or very similar ended events.

Past editions share series_id. Similar events match themes/judging closely. Use results with list_winning_submissions and build_winner_playbook.

List publicly announced winners / awarded submissions for a hackathon.

Prefer past (ended) editions. Returns award, placement, win_signals, and links.

Build a winning strategy from past editions and similar hackathon winners.

Finds related ended hackathons, mines their winning submissions for patterns (polish bar, tracks, win signals), maps those to the current project, and returns a concrete playbook: positioning, must-ship artifacts, heavy inspo moves from nearest past winners, and a judge checklist.

This is inspiration/strategy — adapt patterns; do not copy code or brands.

List specialized HackScout scout agents (Nia-inspired).

Prefer these for multi-step goals instead of manually chaining many atomic tools. Each agent has scoped capabilities, modes (quick/deep/oracle), and returns phased events + citations + a structured report.

Spawn a specialized scout agent and return its phased, cited report.

Agents:

• opportunity_scout — discover/filter hackathons

• fit_strategist — fit score + track + go/no-go

• gallery_intel — deploy into a public gallery (Document-Agent style)

• winner_coach — past-edition / similar winner playbook

• submission_advisor — Oracle-style multi-phase win plan

Modes (Nia research ladder):

• quick — minimal phases

• deep — richer sources / more neighbors

• oracle — full DISCOVER→FIT→WINNERS→GALLERY→SYNTHESIZE (submission_advisor)

Returns run_id, events, citations, and report. Retrieve later via get_scout_run.

Fetch a previous scout agent run by run_id (job-style retrieval).

Add a hackathon to the watchlist and store an initial gallery snapshot.

Refresh the stored gallery snapshot for a watched (or any) hackathon.

Note: live re-extraction requires calling extract_public_submissions first when TINYFISH_API_KEY is configured.

Diff current public submissions against the last watch snapshot.

since_timestamp is accepted for API compatibility; MVP diffs against the stored watch snapshot (or creates one if missing).

Check watched hackathons for landscape changes that may affect a project.

Returns go/no-go style alerts when crowding or gallery composition shifts.