0.0 / 30
What changed in the harness
Selection accuracy 98→98, token cost up 12%, unconfirmed writes 0%→0%.
Category breakdown
Where the score comes from.
Earned points across the four signals Gradable measures. Safety and Legibility are scored out of 30; Economics and Discoverability are scored out of 20.
01Safety
02Legibility
24.2 / 30
03Economics
19.7 / 20
04Discoverability
14.0 / 20
Highest-impact fix
Estimated gain +30 pointsAdd explicit identity and permission preflight tools
Expose machine-readable principal/tenant confirmation and a non-mutating permission check so agents can verify both before destructive actions.
Description evidence
Defects and rewrites.
13 defects found across the exposed tool descriptions. Suggested rewrites make purpose, inputs, boundaries, and returns easier for an agent to understand.
| Tool | Defect types | Suggested rewrite |
|---|---|---|
get_events |
no_return_description |
Returns real-time business events — funding rounds, acquisitions, executive hires, government contracts, and product launches — matching your filters. Filter by event types (comma-separated: funding, acquisition, hiring, contract, product_launch), look-back days (default 7), industries, locations, and result limit (default 50, max 200). |
get_market_pulse |
no_return_description |
Returns a real-time market activity overview: funding totals, acquisition counts, executive moves, contracts, and product launches aggregated for the past 7 and 30 days, so you can gauge market momentum at a glance. |
get_market_brief |
no_return_description |
Returns today's AI-generated strategic intelligence brief: a narrative analysis of the most important market movements, patterns, and opportunities, synthesized daily to inform strategic decisions. |
manage_watchlist |
no_return_description |
Add, remove, or list companies on your watchlist; tracked companies generate alerts when they have new events. Returns the outcome of the requested action — your tracked companies for 'list', and confirmation of the change for 'add'/'remove' using the given company domains. |
get_watchlist_events |
name_restates_behavior no_return_description |
Returns recent events — funding, acquisitions, executive hires, and contracts — for the companies you track on your watchlist, so you can react to new developments at monitored accounts. Optionally filter by comma-separated event types ('types') and look-back window ('days', default 7). |
get_usage |
no_return_description |
Returns your FundzWatch API usage details — calls made, usage limits, and current tier — so you can track consumption against your plan. |
get_funded_and_hiring |
no_return_description |
Returns companies with a verified funding round in the last 12 months AND live hiring evidence (open ATS roles, executive hires, hiring signals) — the strongest buying window — score-ranked and refreshed daily. Optionally filter by company name ('q') or 2-letter US state ('state'). No API key required. |
get_refinancing_windows |
no_return_description |
Returns companies whose active UCC-1 liens lapse within 12 months, each with the dated refinancing window (Renewal Radar). UCC liens expire after 5 years — the incumbent lender must re-file or lose priority — so each approaching lapse is a dated refinancing opportunity for lenders, MCA/ISO, and equipment finance. Optionally filter by company name ('q') or 2-letter US state ('state'). California + Colorado coverage (CO refreshed daily). No API key required. |
get_stacked_borrowers |
no_return_description |
Returns companies with active secured debt (UCC-1) from 2+ distinct lenders and a filing in the last 24 months — proven appetite for layered financing — making them second-position and refi/consolidation targets. Optionally filter by company name ('q') or 2-letter US state ('state'). California + Colorado coverage (CO refreshed daily). No API key required. |
get_benefit_plans_in_play |
no_return_description |
Returns recently funded companies (≤24 months) that have a DOL Form 5500 benefit-plan filing on record, with plan renewal timing, participant count vs current headcount ('outgrowing their plan'), and the incumbent insurance carrier from Schedule A. For benefits brokers, retirement-plan advisors, PEOs, and HR-tech sellers. Optionally filter by company name ('q') or 2-letter US state ('state'). No API key required. |
get_lender_directory |
no_return_description |
Returns who finances whom: a ranked directory of 8,600+ UCC secured parties (lenders) with each lender's filing volume, active-lien counts, and lapsing-soon exposure. Search by lender name ('q', e.g. 'Wells Fargo') and page through results (50/page, default 1). California + Colorado coverage (CO refreshed daily). No API key required. |
get_money_in_motion |
no_return_description |
Returns companies with an executive-move announcement in the last 180 days AND a verified funding round in the last 24 months — the wealth advisor's 'money in motion' moment: a new comp package (often equity), an old 401(k) to roll over, benefits about to be re-evaluated. Score-ranked and refreshed daily. For RIAs, wealth managers, retirement-plan advisors, and executive-benefits sellers. Optionally filter by company name ('q') or 2-letter US state ('state'). No API key required. |
get_broker_directory |
no_return_description |
Returns who broker-of-records whom: a ranked directory of 65,000+ benefits brokers from DOL Form 5500 Schedule A Part 1, with each broker's filings carried, sponsor counts, and commission volume. For carriers, broker M&A, benefits-tech sellers, and brokers scouting competitors. Search by broker name ('q', e.g. 'Lockton'), filter by 2-letter US state ('state'), and page through results (50/page, default 1). No API key required. |
Selection evidence
Confusable tool pairs.
7 pairs where similar names or overlapping descriptions may send an agent toward the wrong tool.
| Tool A | Tool B | Confidence | Why they collide |
|---|---|---|---|
get_market_pulse |
get_market_brief |
high | Both are zero-input market overview tools with overlapping vocabulary ('market', 'get'); a task like 'give me the market overview' or 'summarize what's happening in the market' fits both, and the agent can't disambiguate from arguments since neither takes any. |
get_events |
get_watchlist_events |
medium | A task like 'show me recent events' or 'tell me about new funding and hires' matches get_events, but a user who tracks companies might mean their watchlist; phrasing such as 'events for my companies' or bare 'get events' could route to the generic tool instead of the watchlist-scoped one. |
get_events |
get_market_pulse |
medium | Both surface real-time funding, acquisition, executive, and contract activity; a task like 'what's the latest market activity in fintech' could plausibly hit either the aggregate overview (get_market_pulse) or the raw event feed (get_events). |
get_lender_directory |
get_broker_directory |
medium | Both are semantically parallel 'who finances/brokers whom' directories with ranked volume; a task like 'find the top financing providers' or 'look up the directory of financial intermediaries' is ambiguous between lenders (UCC secured parties) and benefits brokers. |
manage_watchlist |
get_watchlist_events |
medium | A task like 'show me my watchlist' could trigger the 'list' action on manage_watchlist, but could equally mean 'show me what my watchlist companies have been up to' (get_watchlist_events), so the agent may pick the wrong member without clarifying. |
get_funded_and_hiring |
get_money_in_motion |
medium | Both return companies matched on a verified funding round plus a people/hiring signal (hiring evidence vs. executive moves, which overlap); a task like 'companies that recently raised and are hiring executives' could plausibly select either. |
get_refinancing_windows |
get_stacked_borrowers |
low | Both are UCC-1 based, CA/CO, zero-API-key tools targeting financing opportunities; a task like 'find companies that are refinancing or refinancing candidates' could land on the wrong one, though 'expiring liens' vs 'multiple lenders' is described distinctly. |
Compare the field