Where can AI agents find verified skills?
Skills, probed before listed. Skill directories list whatever is uploaded. HIVE402 Skills works like the HIVE402 endpoint catalog — where 153 candidate endpoints were probed and only the 132 that answered a real payment challenge were listed.
Skills, probed before listed. Skill directories list whatever is uploaded. HIVE402 Skills works like the HIVE402 endpoint catalog — where 153 candidate endpoints were probed and only the 132 that answered a real payment challenge were listed. Here the probe is a verification gate, and the operator's own library went through it first: 28 candidate skills enumerated, 15 excluded before listing, 13 passed all four checks. Every listed skill is free and MIT-licensed.
The gate: what gets checked before a skill is listed
Two layers, both applied to Script Master Labs' own skills before anyone else's will be accepted:
- Editorial exclusion. Internal-operations material never reaches the public registry: connector guides bound to private credentials, the internal sales playbook, and notes whose own text says they may be unverified. 15 of the 28 v1 candidates were excluded here, each with its reason recorded in the registry's published verification report.
- Automated checks on the exact file that gets published: (1) it parses — standard frontmatter with a name and a real description; (2) a safety scan finds no credential values, wallet addresses, personal data, destructive shell patterns, or exfiltration-shaped instructions; (3) provenance — author and version are recorded in the file; (4) it is self-contained — every referenced file resolves inside the skill folder, no internal machine paths remain.
The machine-readable feed for agents is skills.json. To install any skill below, fetch its SKILL.md URL and place it in your agent's skills directory.
The v1 registry: 13 verified skills, all free
| Skill | Version | Price | Install (raw SKILL.md) |
|---|---|---|---|
| Answer the Asked Question in Its Context Answer the user exact question in the context they named, by re-reading their exact words and any short reply before answering or confirming closure. | 1.0.0 | Free | answer_asked_question_in_context/SKILL.md |
| Bulk browser batch execution Run a large browser only batch such as a pin or listing swap in direct checkpointed phases, for batches too large for one browser pass or a delegated relay chain. | 1.0.0 | Free | bulk_browser_batch_execution/SKILL.md |
| Decide Under Standing Authorization Decide every resolvable choice on handed over work without presenting decision widgets or dripping owner homework, and bundle genuine owner only inputs into one list in the completion report. | 1.0.0 | Free | decide_under_standing_authorization/SKILL.md |
| Full Scope First Sizing Size builds, scans, listings and cadences at the full scope the work actually needs first, and shrink only on a measured limit, so the user never has to push back against a conservative first pass. | 1.0.0 | Free | full_scope_first_sizing/SKILL.md |
| Ground Attribution and State Claims Verify who actually said, decided or set something, what state handed over work is actually in, and who actually received a message, from primary records before stating any of them as fact. | 1.0.0 | Free | ground_attribution_and_state_claims/SKILL.md |
| Instagram Publish Quota Publish to a shared Instagram account without exhausting its daily API cap, by counting the day's posts across all jobs, ordering the queue, and deferring cleanly on a 429. | 1.0.0 | Free | instagram_publish_quota/SKILL.md |
| Paginated Config Replace Guard Replace a service full configuration safely by fetching every page of the current set, backing it up and diffing key names before and after, so a truncated list can never delete live settings. | 1.0.0 | Free | paginated_config_replace_guard/SKILL.md |
| Plain-First Status Reporting Report finished or in-progress work to the user so the first sentence alone gives the plain verdict, what it means, and any step that is theirs, with receipts and detail after it. | 1.0.0 | Free | plain_first_status_reporting/SKILL.md |
| Preflight a Surface Before Submission Run a per surface compliance and discoverability preflight before filing any server, plugin or endpoint with a venue, directory or registry. | 1.0.0 | Free | preflight_surface_before_submission/SKILL.md |
| Prove Live Before Calling It Wired Classify and prove a service as reachable, wired or settle proven with outside probes and a smallest real settle before reporting that it can take payments. | 1.0.0 | Free | prove_live_before_calling_wired/SKILL.md |
| Unified Hand-Posted Pack Corrects social publishing for pieces the user posts by hand, producing one identical post that works on X, LinkedIn and Threads instead of separate per-channel blocks. | 1.0.0 | Free | unified_hand_posted_pack/SKILL.md |
| Verify auth state before reporting blocked Verify the live sign-in state of a service in the browser and the saved credential store before reporting it as login gated, expired or not connected, and run single-prompt discipline at verification gates. | 1.0.0 | Free | verify_auth_state_before_reporting_blocked/SKILL.md |
| Verify Catalog Price Before Listing Corrects store operations so a price from builder paperwork or a pasted memo is never carried into a listing or launch run until the live catalog record confirms it. | 1.0.0 | Free | verify_catalog_price_before_listing/SKILL.md |
Submissions: how third-party skills will get in
Submissions open in v2 on the same terms the operator's own skills just met: send a skill, it is pulled into the registry workspace, the editorial layer and the automated gate run on it, and it is listed only if it passes — with the per-skill result published either way. That is the endpoint doctrine applied to skills: the listing test is the same one we run on ourselves.
More from Script Master Labs
- HIVE402 — the payability-probed catalog of 132 paid agent APIs ($0.001–$5.00 per call), with MCP search, checkout and Missions.
- OMNIGATE — the x402 gateway and facilitator those payments settle through (Base and Arc; fee leg 0.005 USDC when attached, waived inside the published budget).
- Free x402 Starter Kit — eight MIT templates for building payable agent endpoints.
- The full ScriptMasterLabs catalog — 150+ products: agent APIs, trading systems, books and business kits.