# ShippedByAgents > A sourced database of real accounts of AI coding agents shipping software, with timelines, costs, codebase type and caveats. The database has 163 stories. Each one links to its original source, and every number is quoted from that source; a blank means the source didn't say. Each story is labelled by who published it: vendor (the company that makes the tool, 39 stories), builder (the team or person who did the work, 112) or independent (third-party research, 12). Vendor accounts tend to report bigger gains than independent studies. ## Data - [stories.json](https://www.shippedbyagents.dev/stories.json): every published story as JSON, with source, source type, date (the source publish date, not when we added the story; null if the source gave none; newest-first order and Latest story use this date), tools, models, work type, codebase, key numbers, claimed benefits (as stated by the source), technique tags and caveats - [Query API](https://www.shippedbyagents.dev/api/openapi.json): read-only. GET https://www.shippedbyagents.dev/api/stories filters by work_type, codebase, tool, source_type, has_cost and has_duration (up to 25 results); GET https://www.shippedbyagents.dev/api/stories/{slug} returns one story. No API key. Send an X-Identity header with a contact URL or email and a User-Agent naming your client. Limits: 30 requests a minute and 500 a day per identity, 60 a minute per IP. The OpenAPI document needs no identity - Typed fields: duration_days, cost_usd and prs_merged are null unless the source stated a clear number; prose numbers stay in key numbers. Figures are per-story quotes and are not comparable; we don't publish an average productivity gain - [Stats](https://www.shippedbyagents.dev/stats/): source, tool, work-type, codebase and technique counts, with definitions and limits - [Tools](https://www.shippedbyagents.dev/tools/): stories grouped by coding agent - [Topics](https://www.shippedbyagents.dev/topics/): stories grouped by type of work and codebase - [Techniques](https://www.shippedbyagents.dev/techniques/): stories grouped by how the work was done, in six stages. 148 of 163 stories map to at least one technique - [Method](https://www.shippedbyagents.dev/about): how stories are chosen and labelled - [Submit a story](https://www.shippedbyagents.dev/submit): send a sourced account we're missing - [Sitemap](https://www.shippedbyagents.dev/sitemap-index.xml) ## Tool pages - [Claude Code](https://www.shippedbyagents.dev/tools/claude-code/): 78 stories - [OpenAI Codex](https://www.shippedbyagents.dev/tools/codex/): 28 stories - [Cursor](https://www.shippedbyagents.dev/tools/cursor/): 19 stories - [GitHub Copilot](https://www.shippedbyagents.dev/tools/github-copilot/): 12 stories - [Claude](https://www.shippedbyagents.dev/tools/claude/): 9 stories - [Devin](https://www.shippedbyagents.dev/tools/devin/): 6 stories - [OpenCode](https://www.shippedbyagents.dev/tools/opencode/): 6 stories - [Claude Agent SDK](https://www.shippedbyagents.dev/tools/claude-agent-sdk/): 3 stories - [Amazon Q Developer](https://www.shippedbyagents.dev/tools/amazon-q-developer/): 3 stories - [Unstated](https://www.shippedbyagents.dev/tools/unstated/): 30 stories ## Topic pages - [Language ports](https://www.shippedbyagents.dev/topics/language-ports/): 14 stories. Moving a codebase from one programming language to another, such as Scala to Java or Zig to Rust. - [Test migrations](https://www.shippedbyagents.dev/topics/test-migrations/): 4 stories. Moving a test suite from one framework to another, such as Enzyme to React Testing Library or JUnit 3 to JUnit 4. - [Migrations](https://www.shippedbyagents.dev/topics/migrations/): 25 stories. Moving existing code to a new library, platform or architecture, often as many pull requests. - [Upgrades and modernisation](https://www.shippedbyagents.dev/topics/upgrades-and-modernisation/): 14 stories. Version upgrades and bringing old code up to date. - [Refactoring and maintenance](https://www.shippedbyagents.dev/topics/refactor-and-maintenance/): 14 stories. Cleaning up, restructuring or fixing existing code without changing what it does. - [New products built with agents](https://www.shippedbyagents.dev/topics/new-products/): 45 stories. Apps, libraries, games, sites and internal platforms built mostly by coding agents. - [Research and benchmarks](https://www.shippedbyagents.dev/topics/research-and-benchmarks/): 13 stories. Studies and benchmarks that measure what agents do. - [Company rollouts and cost](https://www.shippedbyagents.dev/topics/company-rollouts/): 28 stories. Companies giving agents to many engineers, and how they handle adoption and spend. - [Solo builders](https://www.shippedbyagents.dev/topics/solo-builders/): 8 stories. One person working with agents, outside a company team. - [Brownfield codebases](https://www.shippedbyagents.dev/topics/brownfield/): 104 stories. Agents working in existing codebases rather than starting from scratch. - [Greenfield projects](https://www.shippedbyagents.dev/topics/greenfield/): 44 stories. Agents building new code from an empty repository. - [Independent sources](https://www.shippedbyagents.dev/topics/independent-sources/): 12 stories. Stories from researchers or press with no stake in the tool or the project. - [Vendor case studies](https://www.shippedbyagents.dev/topics/vendor-case-studies/): 39 stories. Stories published by the company selling the tool, or a company it owns. Read the caveats on these closely. ## Technique pages - [Rules, skills and context files](https://www.shippedbyagents.dev/techniques/rules-skills-and-context-files/): 65 stories, in the Context stage. The agent is steered with written instructions and context: AGENTS.md/CLAUDE.md, skill files, rule docs, migration guides, golden examples, indexed code or telemetry. - [Spec or plan agreed first](https://www.shippedbyagents.dev/techniques/spec-or-plan-agreed-first/): 48 stories, in the Context stage. A written spec, plan, design doc or brief is produced (often by the agent from old code) and settled or approved by people before implementation starts. - [Small, bounded batches](https://www.shippedbyagents.dev/techniques/small-bounded-batches/): 37 stories, in the Doing the work stage. Work is cut into small PRs, components, phases or batches with capped size or volume, instead of one big-bang change. - [Parallel agents on isolated copies](https://www.shippedbyagents.dev/techniques/parallel-agents-on-isolated-copies/): 32 stories, in the Doing the work stage. Several agents or sessions work at once, each in its own worktree, container, clone or branch, coordinated through git or tickets. - [Orchestrator and worker agents](https://www.shippedbyagents.dev/techniques/orchestrator-and-worker-agents/): 27 stories, in the Doing the work stage. A lead, planner or orchestrator agent splits the work and hands pieces to worker or specialist agents, sometimes with a judge. - [Deterministic tools plus LLM](https://www.shippedbyagents.dev/techniques/deterministic-tools-plus-llm/): 20 stories, in the Doing the work stage. Classic tooling (codemods, code search, static analysis, build graphs, scripted pipelines) does the mechanical or discovery part and the LLM handles bounded steps. - [Background agents on in-house infrastructure](https://www.shippedbyagents.dev/techniques/background-agents-on-in-house-infrastructure/): 26 stories, in the Doing the work stage. Agents run in cloud sandboxes or on internal platforms, triggered from tickets, Slack, CI or schedules rather than a developer's laptop. - [Automated verification](https://www.shippedbyagents.dev/techniques/automated-verification/): 83 stories, in the Verification stage. The agent's work is checked automatically: tests, compilers, linters, CI gates or fix loops, including parity harnesses, coverage or eval sets built before the change. - [Human review](https://www.shippedbyagents.dev/techniques/human-review/): 38 stories, in the Review and merge stage. A person reviews or approves the change (or the plan) before merge, or the owning team makes the final call. - [Agent review](https://www.shippedbyagents.dev/techniques/agent-review/): 36 stories, in the Review and merge stage. A separate agent or a different model reviews, audits or tries to break each change. - [Risk-tiered routing](https://www.shippedbyagents.dev/techniques/risk-tiered-routing/): 9 stories, in the Review and merge stage. Rules that route each change by risk: low-risk changes auto-merge or get agent approval, risky areas go to a human. - [Staged rollout of shipped code](https://www.shippedbyagents.dev/techniques/staged-rollout-of-shipped-code/): 7 stories, in the Release stage. The shipped code goes out gradually: feature flag, canary with rollback, pre-release, closed testing or a rehearsed data migration. - [Enablement, spend controls and tool pilots](https://www.shippedbyagents.dev/techniques/enablement-spend-controls-and-tool-pilots/): 17 stories, in the Organisation stage. Company-level moves: preinstalled or self-hosted setups, peer and champion adoption, budgets and token policies, and pilots or proofs of concept of the tool before wider rollout.