{"name":"Agent Bridge (devnet)","description":"A team of engineering agents. Send a message and it is routed to whichever member fits; address one directly at its own endpoint to choose.","version":"0.1.0","supportedInterfaces":[{"url":"https://agent-bridge.compute.dev.ddcdragon.com/a2a","protocolBinding":"JSONRPC","protocolVersion":"1.0"}],"capabilities":{"streaming":true,"pushNotifications":false,"extendedAgentCard":false,"extensions":[{"uri":"https://github.com/a2aproject/a2a-samples/extensions/traceability/v1","description":"Every tool call and agent call in the turn as a ResponseTrace on the status message: tool name, arguments, start and end time, and latency in MILLISECONDS. Anything policy refused carries a 'refused' entry in additionalAttributes, and anything that failed an 'error' entry, since the schema has no field for either. Cost and token counts are omitted: usage is metered per turn, not per step, and is on Task.metadata.","required":false}]},"securitySchemes":{"bearer":{"httpAuthSecurityScheme":{"scheme":"Bearer","description":"A service-to-service key issued by this deployment's operator. It authenticates the caller and carries no identity: whose work a call is for travels in the message's `metadata.onBehalfOf` -- the 0x-hex wallet public key of the person the platform verified at publish time -- and is read only on a call this key has already authenticated."}}},"securityRequirements":[{"schemes":{"bearer":{"list":[]}}}],"defaultInputModes":["text/plain","text/markdown"],"defaultOutputModes":["text/plain","text/markdown","text/html","application/json"],"skills":[{"id":"abc","name":"abc","description":"x","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"auditor","name":"auditor","description":"Reviews a change for security and correctness — reads the diff and the code it lands in, traces every finding to the source that proves it, and reports what it confirmed separately from what it suspects","tags":["read-only"],"examples":[]},{"id":"calculator","name":"calculator","description":"Calculates","tags":["read-only"],"examples":[]},{"id":"calculator-v2","name":"calculator-v2","description":"Sum number","tags":["read-only"],"examples":[]},{"id":"campaign-analyst","name":"campaign-analyst","description":"Analyzes attributed completed WF1 results and campaign activity; proposes evidence-backed targeting and screening tests without rescoring candidates or changing campaigns.","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"campaign-census","name":"campaign-census","description":"Counts how many HeyReach outreach campaigns are currently running.","tags":["read-only"],"examples":[]},{"id":"campaign-manager","name":"campaign-manager","description":"Manages Campaigns","tags":["read-only"],"examples":[]},{"id":"campaign-review","name":"campaign-review","description":"HR WF2 coordinator: consumes verified WF1 results and campaign evidence, calls campaign analysis and independent review, and saves next-batch campaign and screening proposals.","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"candidate-reviewer","name":"candidate-reviewer","description":"Reviews a candidate against our hiring bar and scores them.","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"claim-chart","name":"claim-chart","description":"Builds evidence-grounded civil and patent claim charts from matter records, pleadings and supplied authorities; identifies support, contradictions and missing proof","tags":["read-only"],"examples":[]},{"id":"cloud-specialist-reviewer","name":"cloud-specialist-reviewer","description":"Reviews a code diff for how it behaves against remote services — timeouts, retries, fan-out, pagination, cost and blast radius.","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"content","name":"content","description":"Drafts outward-facing content — narrative, scripts, posts, landing copy, FAQs — from approved claims, and audits published copy against them","tags":["read-only","Bash","Glob","Grep","Read","WebFetch"],"examples":[]},{"id":"content-workflow","name":"content-workflow","description":"Coordinates source-grounded research, content drafting, independent review and a saved publisher handoff; publication remains a named human decision. Uses supplied evidence only.","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"data-specialist-reviewer","name":"data-specialist-reviewer","description":"Reviews a code diff for data correctness — schemas, migrations, queries, and anything that can silently corrupt or lose records.","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"ddc-funding-signer","name":"ddc-funding-signer","description":"Sends DDC customer top-ups through the funding middleware and reports per account what the response actually said - either executing a run a forecaster already decided, or carrying out a top-up a person has asked for directly with the account and amount they named","tags":["read-only","Bash","Read"],"examples":[]},{"id":"ddc-usage-analyst","name":"ddc-usage-analyst","description":"Compares DDC usage across every account in its mapping - what each spent, on what, and how many hours of deposit each has left - publishes an HTML report as an artifact, and hands every run to ddc-usage-forecaster, which decides whether anything needs funding","tags":["read-only","Bash","Read"],"examples":[]},{"id":"ddc-usage-forecaster","name":"ddc-usage-forecaster","description":"Judges whether recent DDC usage is representative enough to fund against, decides a top-up per account for the next 25 eras, stages it at the middleware so the addresses never travel through a message, and hands the signer a run id","tags":["read-only","Bash","Read"],"examples":[]},{"id":"demo","name":"demo","description":"Abc","tags":["read-only"],"examples":[]},{"id":"developer","name":"developer","description":"Writes the fix and opens a pull request — code changes, with tests","tags":["author","Bash","Edit","Glob","Grep","Read","Write"],"examples":[]},{"id":"devops","name":"devops","description":"Changes infrastructure — deploys, rollbacks, config and manifest edits","tags":["operator","Bash","Edit","Glob","Grep","Read","Write"],"examples":[]},{"id":"dragon1-sre","name":"dragon1-sre","description":"Reports on and investigates the Cere Network layers on Dragon 1 — inference and the agent layer","tags":["read-only","Bash","Glob","Grep","Read","WebFetch"],"examples":[]},{"id":"gtm","name":"gtm","description":"Coordinates supplied campaign and application evidence into a reviewed sales handoff. Read-only, supplied evidence only; no Notion, Gmail, Slack, Calendar or other connector requests.","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"hello-world","name":"hello-world","description":"say \"hello world","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"hello-world-v2","name":"hello-world-v2","description":"says \"hello world","tags":["read-only","Bash","Grep","Read"],"examples":[]},{"id":"hr","name":"hr","description":"Reviews a candidate record against the criteria for a role — what the evidence supports, what it does not, and what is missing. Reads the record and the role, never the person.","tags":["read-only"],"examples":[]},{"id":"infra-specialist-reviewer","name":"infra-specialist-reviewer","description":"Reviews a code diff for infrastructure correctness — CI, containers, deploy config, and config keys nothing reads.","tags":["read-only"],"examples":[]},{"id":"legal-workflow","name":"legal-workflow","description":"Coordinates a supplied matter’s evidence, reviewed briefing and proposed next actions; keeps legal conclusions and agreement execution with the designated human reviewer. No direct connectors.","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"new-demo-agent","name":"new-demo-agent","description":"Demos capabilities","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"people-ops","name":"people-ops","description":"Turns a completed candidate review into the next action for a human — who has to decide what, by when, and what they need in front of them.","tags":["read-only","Glob","Grep","Read"],"examples":[]},{"id":"pr-review-poster","name":"pr-review-poster","description":"Posts the finished code review onto the pull request on GitHub.","tags":["read-only"],"examples":[]},{"id":"product","name":"product","description":"Answers product questions and drafts product artefacts — specs, decision records, roadmap positions — from the memory bank, and reports where what we ship and what we claim disagree","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"product-manager","name":"product-manager","description":"Turns customer and campaign feedback into evidence-grounded product decisions, change briefs, and human-approved handoffs to the roles that execute them","tags":["read-only","Bash","Glob","Grep","Read","WebFetch"],"examples":[]},{"id":"research","name":"research","description":"Builds evidence-grounded briefs on markets, competitors, partners, accounts and candidates from public sources and the memory bank, with every claim cited and dated","tags":["read-only","WebFetch"],"examples":[]},{"id":"review-judge","name":"review-judge","description":"Approves or rejects the change on the classified risk, and states what must change before it can pass.","tags":["read-only"],"examples":[]},{"id":"risk-classifier","name":"risk-classifier","description":"Merges the four specialist reviews into one deduplicated, ranked risk verdict with a merge gate.","tags":["read-only"],"examples":[]},{"id":"roc-demo-forecaster","name":"roc-demo-forecaster","description":"Forecasts DDC storage usage for the next 30 days from recent history. Created from ROC to prove the workflow builder can register an external agent.","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"rocdeveloper","name":"rocdeveloper","description":"123","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"sales","name":"sales","description":"Reviews supplied evidence for one customer and deal, coordinates a source-checked next-step recommendation, and saves a human-reviewable handoff. No account connectors or external actions.","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"security-specialist-reviewer","name":"security-specialist-reviewer","description":"Reviews a code diff for exploitable weaknesses — missing authz, injection, secrets, unsafe deserialization and loosened controls.","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"sentinel","name":"sentinel","description":"Watches a pull request's checks to a verdict and separates a real failure from a flake or an infrastructure fault, so a human is woken only for the first","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"spec","name":"spec","description":"Turns a goal into a written spec grounded in the memory bank and the code, gets it approved, then a product review before it is done","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"sre","name":"sre","description":"Investigates production incidents from logs, metrics, cluster state and source","tags":["read-only","Bash","Glob","Grep","Read","WebFetch"],"examples":[]},{"id":"system-architect","name":"system-architect","description":"Architect the systems.","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"system-architect-v2","name":"system-architect-v2","description":"Designs the systems","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"system-architector","name":"system-architector","description":"Design new system components and features","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"ui-ux","name":"ui-ux","description":"Owns UI/UX across marketplace, vault and ROC -- answers what these surfaces actually do, checks a claim against them, and checks them against our own standards","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"uladtest","name":"uladtest","description":"xx","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"web-publisher","name":"web-publisher","description":"Manages content-focused websites and landing pages — audits and prepares content changes, opens pull requests, and verifies staging evidence","tags":["author","Bash","Edit","Glob","Grep","Read","WebFetch","Write"],"examples":[]},{"id":"wf1-candidate-review","name":"wf1-candidate-review","description":"HR WF1 coordinator: reviews one permitted candidate and role through real specialist calls, saves a cited scorecard and exact WF2 packet, and prepares the human decision.","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"workflow-reviewer","name":"workflow-reviewer","description":"Independently checks business-workflow reports and saved packets against source evidence, attribution, calculations and actual handoff receipts; returns corrections or a bounded review verdict.","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]},{"id":"zzz","name":"zzz","description":"Architect the systems.","tags":["read-only","Bash","Glob","Grep","Read"],"examples":[]}]}