{"id":"90ca3767a1124ea4f154dece6803af28863c5dc9f664b964cb904f1ec571e285","title":"Microsoft Research presents PDD Kit for verifying agent-written Rust apps","argument":"Microsoft Research describes PDD Kit, a research prototype for Proof-Driven Development (PDD), in which a team reviews what software should do and Verus checks whether specified Rust code meets that specification. The authors argue that this changes the review burden as coding agents produce large amounts of code: instead of relying mainly on line-by-line review, teams can review intent and have a proof check specified properties for every input.\n\nTheir example is the login logic in Pet Store, a Rust storefront. PDD Kit turns a brief request into a reviewed specification, asks questions about ambiguous requirements, and proves the implementation against the agreed terms. The authors report that the work found existing bugs; they also distinguish the proved code from components that remain trusted and are checked with tests. Database lookups, password checking, token signing and other assumptions sit outside the proof, so the assurance depends on those contracts as well as on the specification capturing the intended behavior.\n\nThe authors’ central point is not that a proof replaces human judgment: people still approve what the software is meant to do and what the proof takes on trust. PDD Kit supports Rust only, and the authors describe proofs as requiring real effort.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The demonstration concerns the login logic in one part of Pet Store; the source does not demonstrate the approach at broader production scale. A proof establishes that code meets its specification, not that the specification fully captures what developers intended. Important components, including database lookups, password checking and token signing, are outside the proof and rely on accepted assumptions with tests as checks. The authors describe PDD Kit as a research prototype that supports Rust only and say proofs still take real effort.","speaker":{"name":"Brenda Potts","roleAtSource":"Article author"},"source":{"title":"Formally verifying agent-written full-stack web apps with PDD Kit - Microsoft Research","publisher":"Microsoft Research","format":"Article","url":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","publishedAt":"2026-11-05T16:00:00+00:00","dateBasis":"Date stated by the publisher."},"evidence":[{"text":"PDD Kit turns a team's intent into a formal specification and proves specified properties of agent-written code for every input, using tests where proof does not reach.","excerpt":"PDD Kit makes the agent keep the plan. It works with you to turn your intent into a formal specification in Verus (opens in new tab), proves the agent’s code meets that spec for every input, and falls back to tests where a proof can’t reach.","url":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","relationship":"Source-stated"},{"text":"In the Pet Store example, the authors say PDD Kit was used to formally verify login logic and uncovered existing bugs.","excerpt":"Verus hardened the code and uncovered existing bugs, proving the absence of login bugs with respect to the specification.","url":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","relationship":"Source-stated"},{"text":"The proof covers only specified properties and code within its boundary; several components remain trusted and tested rather than proved.","excerpt":"The two database lookups, check_password over bcrypt, the clock, token signing, the database error message and the handler’s request unpacking sit outside the proof, each behind an assumed contract the developer accepted.","url":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","relationship":"Source-stated"},{"text":"The authors frame the change as shifting more review from generated code to the team's intent, while retaining human approval of that intent.","excerpt":"Instead of relying mainly on line-by-line review of agent-generated code, a team reviews what it meant, in plain language, and a proof holds the code to it.","url":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","relationship":"Source-stated"},{"text":"PDD Kit is a research prototype and currently verifies Rust only; the authors say they want to learn how far the approach scales.","excerpt":"PDD Kit is a research prototype, and it only verifies Rust today since that’s the language Verus works on.","url":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","relationship":"Source-stated"}],"sourceMedia":{"status":"found","sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","checkedAt":"2026-11-09T05:37:53.638Z","items":[{"type":"image","url":"https://www.microsoft.com/en-us/research/wp-content/uploads/2026/10/PDDKitRecap-BlogHeroFeature-1400x788-1.jpg","sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","attribution":"Microsoft Research","basis":"publisher-embed","availability":"verified","label":"PDDKit decision flowchart on green gradient background","verifiedAt":"2026-11-09T05:37:54.009Z"}],"note":"Media declared by the original publisher; availability checked separately from the editorial reading."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How far can PDD Kit scale beyond the Pet Store login example while keeping the specification, proofs, and trusted components manageable?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Microsoft Research presents PDD Kit for verifying agent-written Rust apps","summary":"Microsoft Research describes PDD Kit, a research prototype for Proof-Driven Development (PDD), in which a team reviews what software should do and Verus checks whether specified Rust code meets that specification. The authors argue that this changes the review burden as coding agents produce large amounts of code: instead of relying mainly on line-by-line review, teams can review intent and have a proof check specified properties for every input.\n\nTheir example is the login logic in Pet Store, a Rust storefront. PDD Kit turns a brief request into a reviewed specification, asks questions about ambiguous requirements, and proves the implementation against the agreed terms. The authors report that the work found existing bugs; they also distinguish the proved code from components that remain trusted and are checked with tests. Database lookups, password checking, token signing and other assumptions sit outside the proof, so the assurance depends on those contracts as well as on the specification capturing the intended behavior.\n\nThe authors’ central point is not that a proof replaces human judgment: people still approve what the software is meant to do and what the proof takes on trust. PDD Kit supports Rust only, and the authors describe proofs as requiring real effort.","question":"How far can PDD Kit scale beyond the Pet Store login example while keeping the specification, proofs, and trusted components manageable?","claims":[{"text":"PDD Kit turns a team's intent into a formal specification and proves specified properties of agent-written code for every input, using tests where proof does not reach.","excerpt":"PDD Kit makes the agent keep the plan. It works with you to turn your intent into a formal specification in Verus (opens in new tab), proves the agent’s code meets that spec for every input, and falls back to tests where a proof can’t reach.","support":"source-stated"},{"text":"In the Pet Store example, the authors say PDD Kit was used to formally verify login logic and uncovered existing bugs.","excerpt":"Verus hardened the code and uncovered existing bugs, proving the absence of login bugs with respect to the specification.","support":"source-stated"},{"text":"The proof covers only specified properties and code within its boundary; several components remain trusted and tested rather than proved.","excerpt":"The two database lookups, check_password over bcrypt, the clock, token signing, the database error message and the handler’s request unpacking sit outside the proof, each behind an assumed contract the developer accepted.","support":"source-stated"},{"text":"The authors frame the change as shifting more review from generated code to the team's intent, while retaining human approval of that intent.","excerpt":"Instead of relying mainly on line-by-line review of agent-generated code, a team reviews what it meant, in plain language, and a proof holds the code to it.","support":"source-stated"},{"text":"PDD Kit is a research prototype and currently verifies Rust only; the authors say they want to learn how far the approach scales.","excerpt":"PDD Kit is a research prototype, and it only verifies Rust today since that’s the language Verus works on.","support":"source-stated"}],"limitations":["The demonstration concerns the login logic in one part of Pet Store; the source does not demonstrate the approach at broader production scale.","A proof establishes that code meets its specification, not that the specification fully captures what developers intended.","Important components, including database lookups, password checking and token signing, are outside the proof and rely on accepted assumptions with tests as checks.","The authors describe PDD Kit as a research prototype that supports Rust only and say proofs still take real effort."],"mediaId":"b28ecfb79591f96a93e5ce4dbf8e96ef6ee80c64e74da1500e0255bd7a8e51b1","correctionReason":null,"id":"90ca3767a1124ea4f154dece6803af28863c5dc9f664b964cb904f1ec571e285","publicationId":"da624cb6-9d5b-43c1-80e4-30be24dbb05c","version":1,"sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","sourceTitle":"Formally verifying agent-written full-stack web apps with PDD Kit - Microsoft Research","publisher":"Microsoft Research","publishedAt":"2026-11-05T16:00:00+00:00","originalPublished":{"value":"2026-11-05T16:00:00+00:00","precision":"instant","basis":"publisher_page"},"firstObservedAt":"2026-11-05T17:09:46.266Z","createdAt":"2026-11-10T17:20:14.759Z","historical":true,"entityIds":["microsoft-research"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"b28ecfb79591f96a93e5ce4dbf8e96ef6ee80c64e74da1500e0255bd7a8e51b1","kind":"article","url":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-09T05:37:54.009Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."},{"id":"c40a9ce83e62c8adb75bbbe6175dc278944153d36207c85fb00d72aad4228b47","kind":"image","url":"https://www.microsoft.com/en-us/research/wp-content/uploads/2026/10/PDDKitRecap-BlogHeroFeature-1400x788-1.jpg","sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","label":"PDDKit decision flowchart on green gradient background","provenance":"publisher_page","verification":{"status":"verified","checkedAt":"2026-11-09T05:37:54.009Z","method":"publisher_association_and_head","contentType":"image/jpeg","cacheUntil":1794202674009},"recommendationReason":"Observed on the acquired publisher source; availability checked."}],"sourceMedia":{"status":"found","sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","checkedAt":"2026-11-09T05:37:53.638Z","items":[{"type":"image","url":"https://www.microsoft.com/en-us/research/wp-content/uploads/2026/10/PDDKitRecap-BlogHeroFeature-1400x788-1.jpg","sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","attribution":"Microsoft Research","basis":"publisher-embed","availability":"verified","label":"PDDKit decision flowchart on green gradient background","verifiedAt":"2026-11-09T05:37:54.009Z"}],"note":"Media declared by the original publisher; availability checked separately from the editorial reading."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"b28ecfb79591f96a93e5ce4dbf8e96ef6ee80c64e74da1500e0255bd7a8e51b1","confidence":0.77,"provider":"typesafe","model":"jev-1.13.0","inputHash":"4d780ccc2a86e911f27f6348d60ce25b5be056405e58f4d103acd36c2e7bacf1","decidedAt":"2026-11-09T05:37:55.726Z"},"provenance":{"textHash":"933a56d655052f0942564af8292f0afb911fad92ab6ea0c396d8a61ab0136082","observationId":"0e7ca8a57959ccd127a7df34b203718aeb2b216990e874b76f2dba40cd930fb3","retrievedAt":"2026-11-09T05:37:53.638Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna; stream-draft-repair-v1","runId":"1e8a5f71-b5c6-41ae-88b7-ea32e40de9a2"},"reviewer":{"model":"gpt-6-luna","runId":"f9883477-b85b-4a1b-8aeb-2116c719a89e"}},"audit":{"status":"passed","reviewedAt":"2026-11-10T17:20:14.759Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Brenda Potts","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Brenda Potts","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Brenda Potts","status":"stated","provenance":{"basis":"article-jsonld","sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","locator":"document/html[1]/head[0]/script[71]@3096.@graph[0].author"},"assertionId":"deefd038b43414e71e1dee30fa1ff913a4163be577c367830f18b14246e48487:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T05:37:53.638Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"deefd038b43414e71e1dee30fa1ff913a4163be577c367830f18b14246e48487","sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","contentSha256":"fffaded8aca5cfd98478fd367a68b7bd00794e2439393315097bf1d2261faf11","textHash":"933a56d655052f0942564af8292f0afb911fad92ab6ea0c396d8a61ab0136082","observedAt":"2026-11-09T05:37:53.638Z"}}],"participationCoverage":{"total":2,"returned":1,"complete":false,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","snapshotId":"deefd038b43414e71e1dee30fa1ff913a4163be577c367830f18b14246e48487","contentSha256":"fffaded8aca5cfd98478fd367a68b7bd00794e2439393315097bf1d2261faf11","textHash":"933a56d655052f0942564af8292f0afb911fad92ab6ea0c396d8a61ab0136082","observedAt":"2026-11-09T05:37:53.638Z","basis":"publisher-page-metadata","sourceLinks":17,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/90ca3767a1124ea4f154dece6803af28863c5dc9f664b964cb904f1ec571e285/source"}},"canonicalUrl":"https://escapemeridian.com/readings/microsoft-research-presents-pdd-kit-for-verifying-agent-written-90ca3767","formats":{"html":"/readings/microsoft-research-presents-pdd-kit-for-verifying-agent-written-90ca3767","markdown":"/readings/microsoft-research-presents-pdd-kit-for-verifying-agent-written-90ca3767.md","json":"/readings/microsoft-research-presents-pdd-kit-for-verifying-agent-written-90ca3767.json"},"attribution":"Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.","maintainedGuideLinks":[],"readerContext":{"version":"reader-context-v1","attribution":"Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.","terms":[],"entities":[{"id":"microsoft-research","name":"Microsoft Research","kind":"organization","context":"Primary technical research with named authors, papers and implementation details.","referenceUrls":["https://www.dwarkesh.com/p/satya-nadella","https://www.dwarkesh.com/p/grant-sanderson","https://www.microsoft.com/en-us/research/blog/proof-based-general-purpose-computing-with-magnetar/","https://www.microsoft.com/en-us/research/"],"path":"/entities/microsoft-research","json":"/entities/microsoft-research.json","markdown":"/entities/microsoft-research.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"90ca3767a1124ea4f154dece6803af28863c5dc9f664b964cb904f1ec571e285","revision":"da624cb6-9d5b-43c1-80e4-30be24dbb05c","revisionKind":"publication-id","title":"Microsoft Research presents PDD Kit for verifying agent-written Rust apps","url":"https://escapemeridian.com/readings/microsoft-research-presents-pdd-kit-for-verifying-agent-written-90ca3767"},"source":{"type":"source","id":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","revision":"933a56d655052f0942564af8292f0afb911fad92ab6ea0c396d8a61ab0136082","revisionKind":"sha256-source-text","title":"Formally verifying agent-written full-stack web apps with PDD Kit - Microsoft Research","url":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/"},"status":"available","items":[],"next":null,"scope":"Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter."},"revisions":[{"id":"90ca3767a1124ea4f154dece6803af28863c5dc9f664b964cb904f1ec571e285","title":"Microsoft Research presents PDD Kit for verifying agent-written Rust apps","argument":"Microsoft Research describes PDD Kit, a research prototype for Proof-Driven Development (PDD), in which a team reviews what software should do and Verus checks whether specified Rust code meets that specification. The authors argue that this changes the review burden as coding agents produce large amounts of code: instead of relying mainly on line-by-line review, teams can review intent and have a proof check specified properties for every input.\n\nTheir example is the login logic in Pet Store, a Rust storefront. PDD Kit turns a brief request into a reviewed specification, asks questions about ambiguous requirements, and proves the implementation against the agreed terms. The authors report that the work found existing bugs; they also distinguish the proved code from components that remain trusted and are checked with tests. Database lookups, password checking, token signing and other assumptions sit outside the proof, so the assurance depends on those contracts as well as on the specification capturing the intended behavior.\n\nThe authors’ central point is not that a proof replaces human judgment: people still approve what the software is meant to do and what the proof takes on trust. PDD Kit supports Rust only, and the authors describe proofs as requiring real effort.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The demonstration concerns the login logic in one part of Pet Store; the source does not demonstrate the approach at broader production scale. A proof establishes that code meets its specification, not that the specification fully captures what developers intended. Important components, including database lookups, password checking and token signing, are outside the proof and rely on accepted assumptions with tests as checks. The authors describe PDD Kit as a research prototype that supports Rust only and say proofs still take real effort.","speaker":{"name":"Brenda Potts","roleAtSource":"Article author"},"source":{"title":"Formally verifying agent-written full-stack web apps with PDD Kit - Microsoft Research","publisher":"Microsoft Research","format":"Article","url":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","publishedAt":"2026-11-05T16:00:00+00:00","dateBasis":"Date stated by the publisher."},"evidence":[{"text":"PDD Kit turns a team's intent into a formal specification and proves specified properties of agent-written code for every input, using tests where proof does not reach.","excerpt":"PDD Kit makes the agent keep the plan. It works with you to turn your intent into a formal specification in Verus (opens in new tab), proves the agent’s code meets that spec for every input, and falls back to tests where a proof can’t reach.","url":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","relationship":"Source-stated"},{"text":"In the Pet Store example, the authors say PDD Kit was used to formally verify login logic and uncovered existing bugs.","excerpt":"Verus hardened the code and uncovered existing bugs, proving the absence of login bugs with respect to the specification.","url":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","relationship":"Source-stated"},{"text":"The proof covers only specified properties and code within its boundary; several components remain trusted and tested rather than proved.","excerpt":"The two database lookups, check_password over bcrypt, the clock, token signing, the database error message and the handler’s request unpacking sit outside the proof, each behind an assumed contract the developer accepted.","url":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","relationship":"Source-stated"},{"text":"The authors frame the change as shifting more review from generated code to the team's intent, while retaining human approval of that intent.","excerpt":"Instead of relying mainly on line-by-line review of agent-generated code, a team reviews what it meant, in plain language, and a proof holds the code to it.","url":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","relationship":"Source-stated"},{"text":"PDD Kit is a research prototype and currently verifies Rust only; the authors say they want to learn how far the approach scales.","excerpt":"PDD Kit is a research prototype, and it only verifies Rust today since that’s the language Verus works on.","url":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","relationship":"Source-stated"}],"sourceMedia":{"status":"found","sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","checkedAt":"2026-11-09T05:37:53.638Z","items":[{"type":"image","url":"https://www.microsoft.com/en-us/research/wp-content/uploads/2026/10/PDDKitRecap-BlogHeroFeature-1400x788-1.jpg","sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","attribution":"Microsoft Research","basis":"publisher-embed","availability":"verified","label":"PDDKit decision flowchart on green gradient background","verifiedAt":"2026-11-09T05:37:54.009Z"}],"note":"Media declared by the original publisher; availability checked separately from the editorial reading."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How far can PDD Kit scale beyond the Pet Store login example while keeping the specification, proofs, and trusted components manageable?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Microsoft Research presents PDD Kit for verifying agent-written Rust apps","summary":"Microsoft Research describes PDD Kit, a research prototype for Proof-Driven Development (PDD), in which a team reviews what software should do and Verus checks whether specified Rust code meets that specification. The authors argue that this changes the review burden as coding agents produce large amounts of code: instead of relying mainly on line-by-line review, teams can review intent and have a proof check specified properties for every input.\n\nTheir example is the login logic in Pet Store, a Rust storefront. PDD Kit turns a brief request into a reviewed specification, asks questions about ambiguous requirements, and proves the implementation against the agreed terms. The authors report that the work found existing bugs; they also distinguish the proved code from components that remain trusted and are checked with tests. Database lookups, password checking, token signing and other assumptions sit outside the proof, so the assurance depends on those contracts as well as on the specification capturing the intended behavior.\n\nThe authors’ central point is not that a proof replaces human judgment: people still approve what the software is meant to do and what the proof takes on trust. PDD Kit supports Rust only, and the authors describe proofs as requiring real effort.","question":"How far can PDD Kit scale beyond the Pet Store login example while keeping the specification, proofs, and trusted components manageable?","claims":[{"text":"PDD Kit turns a team's intent into a formal specification and proves specified properties of agent-written code for every input, using tests where proof does not reach.","excerpt":"PDD Kit makes the agent keep the plan. It works with you to turn your intent into a formal specification in Verus (opens in new tab), proves the agent’s code meets that spec for every input, and falls back to tests where a proof can’t reach.","support":"source-stated"},{"text":"In the Pet Store example, the authors say PDD Kit was used to formally verify login logic and uncovered existing bugs.","excerpt":"Verus hardened the code and uncovered existing bugs, proving the absence of login bugs with respect to the specification.","support":"source-stated"},{"text":"The proof covers only specified properties and code within its boundary; several components remain trusted and tested rather than proved.","excerpt":"The two database lookups, check_password over bcrypt, the clock, token signing, the database error message and the handler’s request unpacking sit outside the proof, each behind an assumed contract the developer accepted.","support":"source-stated"},{"text":"The authors frame the change as shifting more review from generated code to the team's intent, while retaining human approval of that intent.","excerpt":"Instead of relying mainly on line-by-line review of agent-generated code, a team reviews what it meant, in plain language, and a proof holds the code to it.","support":"source-stated"},{"text":"PDD Kit is a research prototype and currently verifies Rust only; the authors say they want to learn how far the approach scales.","excerpt":"PDD Kit is a research prototype, and it only verifies Rust today since that’s the language Verus works on.","support":"source-stated"}],"limitations":["The demonstration concerns the login logic in one part of Pet Store; the source does not demonstrate the approach at broader production scale.","A proof establishes that code meets its specification, not that the specification fully captures what developers intended.","Important components, including database lookups, password checking and token signing, are outside the proof and rely on accepted assumptions with tests as checks.","The authors describe PDD Kit as a research prototype that supports Rust only and say proofs still take real effort."],"mediaId":"b28ecfb79591f96a93e5ce4dbf8e96ef6ee80c64e74da1500e0255bd7a8e51b1","correctionReason":null,"id":"90ca3767a1124ea4f154dece6803af28863c5dc9f664b964cb904f1ec571e285","publicationId":"da624cb6-9d5b-43c1-80e4-30be24dbb05c","version":1,"sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","sourceTitle":"Formally verifying agent-written full-stack web apps with PDD Kit - Microsoft Research","publisher":"Microsoft Research","publishedAt":"2026-11-05T16:00:00+00:00","originalPublished":{"value":"2026-11-05T16:00:00+00:00","precision":"instant","basis":"publisher_page"},"firstObservedAt":"2026-11-05T17:09:46.266Z","createdAt":"2026-11-10T17:20:14.759Z","historical":true,"entityIds":["microsoft-research"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"b28ecfb79591f96a93e5ce4dbf8e96ef6ee80c64e74da1500e0255bd7a8e51b1","kind":"article","url":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-09T05:37:54.009Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."},{"id":"c40a9ce83e62c8adb75bbbe6175dc278944153d36207c85fb00d72aad4228b47","kind":"image","url":"https://www.microsoft.com/en-us/research/wp-content/uploads/2026/10/PDDKitRecap-BlogHeroFeature-1400x788-1.jpg","sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","label":"PDDKit decision flowchart on green gradient background","provenance":"publisher_page","verification":{"status":"verified","checkedAt":"2026-11-09T05:37:54.009Z","method":"publisher_association_and_head","contentType":"image/jpeg","cacheUntil":1794202674009},"recommendationReason":"Observed on the acquired publisher source; availability checked."}],"sourceMedia":{"status":"found","sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","checkedAt":"2026-11-09T05:37:53.638Z","items":[{"type":"image","url":"https://www.microsoft.com/en-us/research/wp-content/uploads/2026/10/PDDKitRecap-BlogHeroFeature-1400x788-1.jpg","sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","attribution":"Microsoft Research","basis":"publisher-embed","availability":"verified","label":"PDDKit decision flowchart on green gradient background","verifiedAt":"2026-11-09T05:37:54.009Z"}],"note":"Media declared by the original publisher; availability checked separately from the editorial reading."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"b28ecfb79591f96a93e5ce4dbf8e96ef6ee80c64e74da1500e0255bd7a8e51b1","confidence":0.77,"provider":"typesafe","model":"jev-1.13.0","inputHash":"4d780ccc2a86e911f27f6348d60ce25b5be056405e58f4d103acd36c2e7bacf1","decidedAt":"2026-11-09T05:37:55.726Z"},"provenance":{"textHash":"933a56d655052f0942564af8292f0afb911fad92ab6ea0c396d8a61ab0136082","observationId":"0e7ca8a57959ccd127a7df34b203718aeb2b216990e874b76f2dba40cd930fb3","retrievedAt":"2026-11-09T05:37:53.638Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna; stream-draft-repair-v1","runId":"1e8a5f71-b5c6-41ae-88b7-ea32e40de9a2"},"reviewer":{"model":"gpt-6-luna","runId":"f9883477-b85b-4a1b-8aeb-2116c719a89e"}},"audit":{"status":"passed","reviewedAt":"2026-11-10T17:20:14.759Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Brenda Potts","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Brenda Potts","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Brenda Potts","status":"stated","provenance":{"basis":"article-jsonld","sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","locator":"document/html[1]/head[0]/script[71]@3096.@graph[0].author"},"assertionId":"deefd038b43414e71e1dee30fa1ff913a4163be577c367830f18b14246e48487:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-09T05:37:53.638Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"deefd038b43414e71e1dee30fa1ff913a4163be577c367830f18b14246e48487","sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","contentSha256":"fffaded8aca5cfd98478fd367a68b7bd00794e2439393315097bf1d2261faf11","textHash":"933a56d655052f0942564af8292f0afb911fad92ab6ea0c396d8a61ab0136082","observedAt":"2026-11-09T05:37:53.638Z"}}],"participationCoverage":{"total":2,"returned":1,"complete":false,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://www.microsoft.com/en-us/research/blog/formally-verifying-agent-written-full-stack-web-apps-with-pdd-kit/","snapshotId":"deefd038b43414e71e1dee30fa1ff913a4163be577c367830f18b14246e48487","contentSha256":"fffaded8aca5cfd98478fd367a68b7bd00794e2439393315097bf1d2261faf11","textHash":"933a56d655052f0942564af8292f0afb911fad92ab6ea0c396d8a61ab0136082","observedAt":"2026-11-09T05:37:53.638Z","basis":"publisher-page-metadata","sourceLinks":17,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/90ca3767a1124ea4f154dece6803af28863c5dc9f664b964cb904f1ec571e285/source"}}}],"audits":[{"publicationId":"da624cb6-9d5b-43c1-80e4-30be24dbb05c","verdict":"passed","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading attributes PDD Kit and its claims to Microsoft Research and its authors. Claims such as finding existing bugs and shifting review toward intent stay framed as the authors' reports and arguments, not as independent facts.","excerpt":"This work is a collaboration between the Microsoft Research Catalyst Lab, GitHub Next, and the Research in Software Engineering (RiSE) group."},{"dimension":"summary","status":"pass","explanation":"The title, summary and question match the source. A general reader can follow the actor, the problem (agent code is too large to review line by line), the mechanism (review intent, prove code against a spec, test what proof cannot reach) and the caveats. Proof-Driven Development is explained in plain words. Verus is named as the checker. The question reflects the authors' own stated open issue about scaling.","excerpt":"We expect that trusted boundary to shrink significantly over time, so a growing share of the code can be formally verified, and we want to find out how far this scales."},{"dimension":"date","status":"pass","explanation":"originalPublished (2026-11-05T16:00Z, from the publisher page) matches the feed entry. Nothing in the evidence suggests an earlier first publication.","excerpt":null},{"dimension":"support","status":"pass","explanation":"Each claim is source-stated and backed by an exact excerpt. The limitations are faithful to the source: one example, trusted components, spec-versus-intent gap, Rust-only prototype and real proof effort. The reading avoids turning the demonstration into proven real-world reliability. It keeps the authors' caveat that guarantees depend on the trusted pieces.","excerpt":"The guarantees are only as strong as the six trusted pieces, and two carry most of the weight: check_password for every password promise, and the two lookups for every promise about which account a token names."},{"dimension":"media","status":"pass","explanation":"The selected media is the verified direct link to the original Microsoft Research blog post. It is the most direct way for readers to access the source.","excerpt":null}],"createdAt":"2026-11-10T17:20:14.759Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}