Agent tool schema linter
Paste the tools your agent can call, an MCP server's tool list included, or the events your agents send each other, and see which fields let someone else's words reach the model on the other side.
Paste the tools your agent can call, an MCP server's tool list included, or the events your agents send each other, and see which fields let someone else's words reach the model on the other side.
How long one agent task can run before your SDK and loop settings stop it, when the model is silent, trickling or overloaded, and what to set when that is too long.
A completeness check for a state machine. Write each state and what its events do, and see which pairs nobody decided, which states can't be reached or left, and where two rules disagree.
Paste the tools your agent can call, an MCP server's tool list included, or the events your agents send each other, and see which fields let someone else's words reach the model on the other side.
How long one agent task can run before your SDK and loop settings stop it, when the model is silent, trickling or overloaded, and what to set when that is too long.
A completeness check for a state machine. Write each state and what its events do, and see which pairs nobody decided, which states can't be reached or left, and where two rules disagree.
Pick a call size and see where a voice call's cost lands for mesh, an SFU, an MCU and a server that is the other peer, with real packet sizes, and what would make it cheaper.
Paste both sides' ICE candidates, as SDP lines, a whole offer, candidate JSON or getStats(): the pairs ICE forms, the ones it drops, the order it tries them, and what to fix.
A minute of voice over a network that delays and loses packets: how much audio the decoder makes up, and which buffer and protection fix it, at what cost in delay and bitrate.
One voice frame is lost. UDP conceals that one frame; TCP recovers it and holds every frame behind it, and each one whose playout time passes while it waits is lost to the listener anyway.
Paste an RTP or RTCP packet as hex and see what it says: each field marked on the bytes, the loss and round trip in its reports, and the NACK, PLI, FIR, REMB and TWCC feedback WebRTC sends.
Paste a WebRTC offer or answer and see every line: what it sets, which phase of the call reads it, and what breaks a rule, with the RFC for each.
Paste a STUN message as hex or a Wireshark hex dump: what it is, what to fix, every attribute with addresses un-XORed, and its integrity checked, TURN long-term credentials included.
A fast link is not a fast transfer. One connection is capped by its window and by packet loss, and on a long path the cap is far below what was paid for.
Pick a call size and see where a voice call's cost lands for mesh, an SFU, an MCU and a server that is the other peer, with real packet sizes, and what would make it cheaper.
Paste both sides' ICE candidates, as SDP lines, a whole offer, candidate JSON or getStats(): the pairs ICE forms, the ones it drops, the order it tries them, and what to fix.
A minute of voice over a network that delays and loses packets: how much audio the decoder makes up, and which buffer and protection fix it, at what cost in delay and bitrate.
One voice frame is lost. UDP conceals that one frame; TCP recovers it and holds every frame behind it, and each one whose playout time passes while it waits is lost to the listener anyway.
Paste an RTP or RTCP packet as hex and see what it says: each field marked on the bytes, the loss and round trip in its reports, and the NACK, PLI, FIR, REMB and TWCC feedback WebRTC sends.
Paste a WebRTC offer or answer and see every line: what it sets, which phase of the call reads it, and what breaks a rule, with the RFC for each.
Paste a STUN message as hex or a Wireshark hex dump: what it is, what to fix, every attribute with addresses un-XORed, and its integrity checked, TURN long-term credentials included.
A fast link is not a fast transfer. One connection is capped by its window and by packet loss, and on a long path the cap is far below what was paid for.
A request fans out to N calls and waits for all of them. See how many requests are slow, and what percentile each call needs to meet your target.
More requests than a service can answer, for a minute. Your workers, traffic, timeout, retries, queue policy and queue size: who is answered in time, who waits for nothing, and does it recover.
List a metric's labels and how many values each takes, pick where it is stored, and see how many time series it really writes, which change saves the most, and what that backend bills it in.
Paste latency histograms from Prometheus, Google Cloud, AWS X-Ray or a log query, or raw latencies. See the service's real p99, how far off it can be, and how wrong the averaged per-host p99 is.
Requests at a rate, work that takes a while, and a target for how many may wait, at all or longer than a limit. Erlang C answers how many servers, shows what one fewer would cost, and draws the cliff.
Retries per layer and how deep the chain goes: how many calls reach the bottom service for one user request when every layer retries what it gets, and the extra load on it while it fails.
A call chain's per-try timeouts and retries, top first: the worst case when the bottom hangs, whether it fits the caller's deadline, and which retries can never finish in time.
A request fans out to N calls and waits for all of them. See how many requests are slow, and what percentile each call needs to meet your target.
More requests than a service can answer, for a minute. Your workers, traffic, timeout, retries, queue policy and queue size: who is answered in time, who waits for nothing, and does it recover.
List a metric's labels and how many values each takes, pick where it is stored, and see how many time series it really writes, which change saves the most, and what that backend bills it in.
Paste latency histograms from Prometheus, Google Cloud, AWS X-Ray or a log query, or raw latencies. See the service's real p99, how far off it can be, and how wrong the averaged per-host p99 is.
Requests at a rate, work that takes a while, and a target for how many may wait, at all or longer than a limit. Erlang C answers how many servers, shows what one fewer would cost, and draws the cliff.
Retries per layer and how deep the chain goes: how many calls reach the bottom service for one user request when every layer retries what it gets, and the extra load on it while it fails.
A call chain's per-try timeouts and retries, top first: the worst case when the bottom hangs, whether it fits the caller's deadline, and which retries can never finish in time.
Paste two copies of a function and each one's git log, and see whether they have been changing together or drifting apart, which commits changed only one, and where the copies differ.
Paste your PRs' sizes and review waits. Get how much the wait grows with size, and what splitting a change into k PRs costs in waiting, reviewed one after another, side by side, or in one sitting.
Paste two copies of a function and each one's git log, and see whether they have been changing together or drifting apart, which commits changed only one, and where the copies differ.
Paste your PRs' sizes and review waits. Get how much the wait grows with size, and what splitting a change into k PRs costs in waiting, reviewed one after another, side by side, or in one sitting.