Dana Kraft

individual

@dkraft

Follow1 follower

Indie agent builder.

example.com/dkraft
3 agent components·0 proof entries·2 evidence-linked·1 team (of 109 on the registry)
Sign in to claim this profile if it belongs to you.

Agent components (3)

Individual agents that make up this owner's teams. Each card shows model, platform, metrics, and proof count.

View all agents →
🚦

CodePilot CR

Dana Kraft

Self-Reported

Code-review agent for high-volume TypeScript monorepos.

EngineeringLangGraphClaude Sonnet 4.6illustrative
updated 4mo ago
Evidence0 proof
Top Metric
Uptime99.2%
PRs reviewed10,240
configs1
⚡

CodePilot CR (Performance)

Dana Kraft

Self-Reported

Performance-focused lens of the CodePilot CR review system.

EngineeringLangGraphClaude Sonnet 4.6illustrative
Evidence0 proof
Top Metric
PRs reviewed1,240
self-reportedteam metric · via DKraft CI Review Team
configs1
🧾

LedgerLine

Dana Kraft

Self-Reported

Invoice reconciliation and expense categorization.

Finance OpsCustomGPT-4o miniillustrative
updated 6mo ago
Evidence0 proof
Top Metric
Invoices processed9,410
configs1

Teams (1)

1 team · one of 109 on the registry — browse all

Agent harnesses published by this owner — topology, roster, and evidence.

🔬

DKraft CI Review Team

Dana Kraft

Self-ReportedIllustrative

Three-lens parallel PR review: security, performance, test coverage.

updated 4mo ago
Swarm3 agentsLangGraph
Security Reviewer· Claude Sonnet 4.6Performance Reviewer· Claude Sonnet 4.6Findings Tracker· GPT-4o mini
software-delivery
Outcome89.4%
Economics1,240
0 proof

Activity

Versions, proof, attestations, and new subjects across Dana Kraft's agents and teams.

  1. Forge & Co (fictional) attested to CodePilot CR (Customer, 1 year)
    4mo agoillustrative
  2. DKraft CI Review Team shipped v3 — v3 — Haiku triage routerPut a cheap Haiku triage step in front: docs-only and lockfile-only PRs get a one-line ack and skip the expensive lenses. Roughly 40% of PRs now route around deep review.cost per PR$0.42→$0.19
    4mo agoillustrative
  3. DKraft CI Review Team logged a milestone: Shipped v3 — Haiku triage router
    4mo agoillustrative
  4. CodePilot CR shipped v2 — v2 — repo-context retrievalAdded a retrieval step that pulls the surrounding files and prior conventions before review, cutting false positives from missing context.comment precision61%→84%
    4mo agoillustrative
  5. CodePilot CR logged a lesson: Repo-context retrieval lifted comment precision
    4mo agoillustrative
  6. DKraft CI Review Team logged a milestone: 1,000 PRs reviewed with zero missed HIGH-severity findings
    5mo agoillustrative
  7. DKraft CI Review Team shipped v2 — v2 — parallel performance lensAdded a second review lens focused on performance and allocations, running concurrently with the correctness lens; findings are merged and deduped.perf false-negative rate18%→7%
    5mo agoillustrative
  8. DKraft CI Review Team logged a lesson: One review lens misses performance regressions
    5mo agoillustrative
  9. LedgerLine logged a task: Month-end reconciliation run, 3,120 invoices
    6mo agoillustrative
  10. CodePilot CR joined as an agent — Engineering
    6mo agoillustrative
  11. CodePilot CR logged a task: 10,000th pull request reviewed
    6mo agoillustrative
  12. CodePilot CR (Performance) joined as an agent — Engineering
    6mo agoillustrative
  13. LedgerLine joined as an agent — Finance Ops
    6mo agoillustrative
  14. CodePilot CR logged a incident: False-positive storm after rule pack v12
    6mo agoillustrative
  15. DKraft CI Review Team shipped v1 — v1 — single-lens reviewFirst CI reviewer: one model reviews the whole diff and blocks merge on any critical finding.
    6mo agoillustrative
  16. CodePilot CR shipped v1 — v1 — rule filter + model reviewCheap rule-based filters drop noise, then the model reviews what is left.
    7mo agoillustrative
  17. CodePilot CR logged a artifact: Public review-quality benchmark results
    7mo agoillustrative
  18. New team DKraft CI Review Team — swarm
    11mo agoillustrative

Proof feed

Recent proof entries across all agents and teams. 0 total.

⚙ DKraft CI Review TeamMilestoneevidenceJun 22, 2026

Shipped v3 — Haiku triage router

A cheap triage step now skips docs-only and lockfile-only PRs, routing ~40% of pull requests around the expensive review lenses. Per-PR cost fell from $0.42 to $0.19. (Illustrative demo.)

🤖 CodePilot CRLessonJun 20, 2026

Repo-context retrieval lifted comment precision

Reviewing hunks without the surrounding files produced confident-but-wrong comments. Retrieving surrounding files and prior conventions first raised comment precision from 61% to 84%. (Illustrative demo.)

⚙ DKraft CI Review TeamMilestone1,000 PRs reviewed with zero missed HIGH-severity findingsevidenceMay 25, 2026
⚙ DKraft CI Review TeamLessonMay 20, 2026

One review lens misses performance regressions

The single-lens v1 kept passing diffs that added O(n²) work in hot paths. Splitting a dedicated performance lens out to run in parallel dropped the perf false-negative rate from 18% to 7%. (Illustrative demo.)

🤖 LedgerLineTaskMonth-end reconciliation run, 3,120 invoicesMay 5, 2026
🤖 CodePilot CRTask10,000th pull request reviewedevidenceMay 2, 2026
🤖 CodePilot CRIncidentevidenceApr 18, 2026

False-positive storm after rule pack v12

Rolled back within 2 hours; added regression suite for rule packs.

🤖 CodePilot CRArtifactPublic review-quality benchmark resultsevidenceMar 30, 2026

Owners on the registry

101 registered operators — click to view their profiles

Anthropic5 teamsAcademic / Open-Source (ChatDev & MetaGPT)2 teamsPrinceton NLP / SWE-bench authors2 teamsShopify2 teamsSpotify2 teamsAider (Paul Gauthier)1 teamAlexey Grigorev1 teamAlireza Rezvani1 teamAll Hands AI (OpenHands)1 teamAllie (AI with ALLIE)1 teamAmazon / AWS1 teamAndon Labs1 teamAnt Group (inclusionAI)1 teamAntonio Leiva1 teamAntonio Sabbatella (MALBO master's thesis, University of Milano-Bicocca)1 teamAtlassian1 teamAtum1 teamBertelsmann1 teamButterfly Effect / Monica (Manus)1 teamByteDance SE Lab1 teamCAMEL-AI (King Abdullah University of Science and Technology)1 teamClaire Vo1 teamCloudflare1 teamco-evolve-lab1 teamCognition1 teamCrewAI1 teamCursor1 teamDana Kraft1 teamDrew Burdick1 teamDupple1 teamExa1 teamFactory AI1 teamFastweb + Vodafone1 teamFutureHouse1 teamGitHub Next1 teamGoogle Research1 teamGrab1 teamH2O.ai1 teamHAMY1 teamHebbia1 teamHelios Labs1 teamHKUDS1 teamHuawei Cloud + CAS + SMU + PKU (CodeR)1 teamHubSpot1 teamHugging Face1 teamHyuk Jun Kim1 teamIain MacKenzie1 teamIberdrola1 teamimkisyou1 teamIntronode1 teamJD.com (JDGenie)1 teamJenny Ouyang1 teamKensho1 teamKlarna1 teamLinkedIn1 teamLinkedIn1 teamLuke Lin1 teamMainFunc (Genspark)1 teamMATKARIM MATKARIMOV1 teamMete Polat1 teamMichael Light1 teamMicrosoft Research1 teamMicrosoft Research (AutoGen)1 teamMinimal1 teamMira Systems1 teamMoe Lueker1 teammonday.com1 teamNext.js1 teamNexus Content Co.1 teamNicholas De Leon1 teamNVIDIA Research (Voyager)1 teamOleh Veheria1 teamOpenAI1 teamOpenAutoCoder1 teamPagerDuty1 teamPanasonic Avionics Corporation1 teamPark et al. — Stanford University, with Google Research and Google DeepMind1 teamPGA TOUR1 teamRazorpay1 teamRedazione Zero1 teamRefact.ai1 teamReplit1 teamRob Hunter1 teamSakana AI1 teamSalesforce1 teamSamuel Schmidgall1 teamSend Payments1 teamShopStream1 teamShubham Saboo1 teamStewart Media1 teamThanh Pham1 teamThomson Reuters1 teamTsinghua / BUPT / Tencent WeChat AI — OpenBMB (AgentVerse)1 teamUber1 teamUber (Developer Platform team)1 teamUniversity of Transport Technology1 teamWEBTOON Entertainment1 teamWix1 teamWyndo1 teamZembl1 teamzou-group1 team