mirror of
https://github.com/rohitg00/ai-engineering-from-scratch.git
synced 2026-10-02 10:04:49 +08:00
Rohit Ghumare
3ecf630b0f
* feat(projects): workflow-hooks stage 1 * feat(projects): workflow-hooks stage 2 * feat(projects): workflow-hooks stage 3 * feat(projects): workflow-hooks stage 4 * fix(projects): preserve sponsor navigation in project footer * feat(projects): agent-budget-planner stage 1 * feat(projects): agent-budget-planner stage 2 * feat(projects): agent-budget-planner stage 3 * feat(projects): agent-budget-planner stage 4 * feat(projects): agent-trace-debugger stage 1 * feat(projects): agent-trace-debugger stage 2 * feat(projects): agent-trace-debugger stage 3 * feat(projects): agent-trace-debugger stage 4 * feat(projects): browser-agent stage 1 * feat(projects): browser-agent stage 2 * feat(projects): browser-agent stage 3 * feat(projects): browser-agent stage 4 * feat(projects): calendar-focus-planner stage 1 * feat(projects): calendar-focus-planner stage 2 * feat(projects): calendar-focus-planner stage 3 * feat(projects): calendar-focus-planner stage 4 * feat(projects): changelog-writer-from-git stage 1 * feat(projects): changelog-writer-from-git stage 2 * feat(projects): changelog-writer-from-git stage 3 * feat(projects): changelog-writer-from-git stage 4 * feat(projects): cloud-agent-with-aws-strands stage 1 * feat(projects): cloud-agent-with-aws-strands stage 2 * feat(projects): cloud-agent-with-aws-strands stage 3 * feat(projects): cloud-agent-with-aws-strands stage 4 * feat(projects): csv-sql-question-workbench stage 1 * feat(projects): csv-sql-question-workbench stage 2 * feat(projects): csv-sql-question-workbench stage 3 * feat(projects): csv-sql-question-workbench stage 4 * feat(projects): dataset-split-auditor stage 1 * feat(projects): dataset-split-auditor stage 2 * feat(projects): dataset-split-auditor stage 3 * feat(projects): dataset-split-auditor stage 4 * feat(projects): desktop-control stage 1 * feat(projects): desktop-control stage 2 * feat(projects): desktop-control stage 3 * feat(projects): desktop-control stage 4 * feat(projects): distributed-eval-farm stage 1 * feat(projects): distributed-eval-farm stage 2 * feat(projects): distributed-eval-farm stage 3 * feat(projects): distributed-eval-farm stage 4 * feat(projects): doc-qa-with-citations stage 1 * feat(projects): doc-qa-with-citations stage 2 * feat(projects): doc-qa-with-citations stage 3 * feat(projects): doc-qa-with-citations stage 4 * feat(projects): document-extraction-desk stage 1 * feat(projects): document-extraction-desk stage 2 * feat(projects): document-extraction-desk stage 3 * feat(projects): document-extraction-desk stage 4 * feat(projects): durable-agent-jobs stage 1 * feat(projects): durable-agent-jobs stage 2 * feat(projects): durable-agent-jobs stage 3 * feat(projects): durable-agent-jobs stage 4 * feat(projects): feedback-theme-board stage 1 * feat(projects): feedback-theme-board stage 2 * feat(projects): feedback-theme-board stage 3 * feat(projects): feedback-theme-board stage 4 * feat(projects): harness-bench stage 1 * feat(projects): harness-bench stage 2 * feat(projects): harness-bench stage 3 * feat(projects): harness-bench stage 4 * feat(projects): inbox-triage-desk stage 1 * feat(projects): inbox-triage-desk stage 2 * feat(projects): inbox-triage-desk stage 3 * feat(projects): inbox-triage-desk stage 4 * feat(projects): json-schema-output-guard stage 1 * feat(projects): json-schema-output-guard stage 2 * feat(projects): json-schema-output-guard stage 3 * feat(projects): json-schema-output-guard stage 4 * feat(projects): llm-gateway-with-fallbacks stage 1 * feat(projects): llm-gateway-with-fallbacks stage 2 * feat(projects): llm-gateway-with-fallbacks stage 3 * feat(projects): llm-gateway-with-fallbacks stage 4 * feat(projects): local-model-eval-harness stage 1 * feat(projects): local-model-eval-harness stage 2 * feat(projects): local-model-eval-harness stage 3 * feat(projects): local-model-eval-harness stage 4 * feat(projects): mcp-at-scale stage 1 * feat(projects): mcp-at-scale stage 2 * feat(projects): mcp-at-scale stage 3 * feat(projects): mcp-at-scale stage 4 * feat(projects): mcp-at-scale stage 5 * feat(projects): meeting-notes-to-actions stage 1 * feat(projects): meeting-notes-to-actions stage 2 * feat(projects): meeting-notes-to-actions stage 3 * feat(projects): meeting-notes-to-actions stage 4 * feat(projects): memory-server stage 1 * feat(projects): memory-server stage 2 * feat(projects): memory-server stage 3 * feat(projects): memory-server stage 4 * feat(projects): multi-agent-code-review-panel stage 1 * feat(projects): multi-agent-code-review-panel stage 2 * feat(projects): multi-agent-code-review-panel stage 3 * feat(projects): multi-agent-code-review-panel stage 4 * feat(projects): postmortem-writer stage 1 * feat(projects): postmortem-writer stage 2 * feat(projects): postmortem-writer stage 3 * feat(projects): postmortem-writer stage 4 * feat(projects): pr-review-reporter stage 1 * feat(projects): pr-review-reporter stage 2 * feat(projects): pr-review-reporter stage 3 * feat(projects): pr-review-reporter stage 4 * feat(projects): prompt-regression-tester stage 1 * feat(projects): prompt-regression-tester stage 2 * feat(projects): prompt-regression-tester stage 3 * feat(projects): prompt-regression-tester stage 4 * feat(projects): rag-freshness-pipeline stage 1 * feat(projects): rag-freshness-pipeline stage 2 * feat(projects): rag-freshness-pipeline stage 3 * feat(projects): rag-freshness-pipeline stage 4 * feat(projects): report-judge stage 1 * feat(projects): report-judge stage 2 * feat(projects): report-judge stage 3 * feat(projects): report-judge stage 4 * feat(projects): research-report-agent stage 1 * feat(projects): research-report-agent stage 2 * feat(projects): research-report-agent stage 3 * feat(projects): research-report-agent stage 4 * feat(projects): research-report-agent stage 5 * feat(projects): research-report-agent stage 6 * feat(projects): research-report-agent stage 7 * feat(projects): retrieval-evaluation-lab stage 1 * feat(projects): retrieval-evaluation-lab stage 2 * feat(projects): retrieval-evaluation-lab stage 3 * feat(projects): retrieval-evaluation-lab stage 4 * feat(projects): rust-agent-shell stage 1 * feat(projects): rust-agent-shell stage 2 * feat(projects): rust-agent-shell stage 3 * feat(projects): rust-agent-shell stage 4 * feat(projects): sandbox-ladder stage 1 * feat(projects): sandbox-ladder stage 2 * feat(projects): sandbox-ladder stage 3 * feat(projects): sandbox-ladder stage 4 * feat(projects): self-improving-skill-loop stage 1 * feat(projects): self-improving-skill-loop stage 2 * feat(projects): self-improving-skill-loop stage 3 * feat(projects): self-improving-skill-loop stage 4 * feat(projects): semantic-notes-search stage 1 * feat(projects): semantic-notes-search stage 2 * feat(projects): semantic-notes-search stage 3 * feat(projects): semantic-notes-search stage 4 * feat(projects): skill-installer stage 1 * feat(projects): skill-installer stage 2 * feat(projects): skill-installer stage 3 * feat(projects): skill-installer stage 4 * feat(projects): skill-router stage 1 * feat(projects): skill-router stage 2 * feat(projects): skill-router stage 3 * feat(projects): skill-router stage 4 * feat(projects): skill-scanner stage 1 * feat(projects): skill-scanner stage 2 * feat(projects): skill-scanner stage 3 * feat(projects): skill-scanner stage 4 * feat(projects): skill-validator stage 1 * feat(projects): skill-validator stage 2 * feat(projects): skill-validator stage 3 * feat(projects): skill-validator stage 4 * feat(projects): source-grounded-study-coach stage 1 * feat(projects): source-grounded-study-coach stage 2 * feat(projects): source-grounded-study-coach stage 3 * feat(projects): source-grounded-study-coach stage 4 * feat(projects): support-agent-with-google-adk stage 1 * feat(projects): support-agent-with-google-adk stage 2 * feat(projects): support-agent-with-google-adk stage 3 * feat(projects): support-agent-with-google-adk stage 4 * feat(projects): tiny-coding-agent stage 1 * feat(projects): tiny-coding-agent stage 2 * feat(projects): tiny-coding-agent stage 3 * feat(projects): tiny-coding-agent stage 4 * feat(projects): token-counter-and-cost-meter stage 1 * feat(projects): token-counter-and-cost-meter stage 2 * feat(projects): token-counter-and-cost-meter stage 3 * feat(projects): token-counter-and-cost-meter stage 4 * feat(projects): tool-call-firewall stage 1 * feat(projects): tool-call-firewall stage 2 * feat(projects): tool-call-firewall stage 3 * feat(projects): tool-call-firewall stage 4 * feat(projects): typed-workflow-agent-with-mastra stage 1 * feat(projects): typed-workflow-agent-with-mastra stage 2 * feat(projects): typed-workflow-agent-with-mastra stage 3 * feat(projects): typed-workflow-agent-with-mastra stage 4 * feat(projects): visual-evidence-library stage 1 * feat(projects): visual-evidence-library stage 2 * feat(projects): visual-evidence-library stage 3 * feat(projects): visual-evidence-library stage 4 * feat(projects): voice-note-transcriber-pipeline stage 1 * feat(projects): voice-note-transcriber-pipeline stage 2 * feat(projects): voice-note-transcriber-pipeline stage 3 * feat(projects): voice-note-transcriber-pipeline stage 4 * feat(projects): web-change-brief stage 1 * feat(projects): web-change-brief stage 2 * feat(projects): web-change-brief stage 3 * feat(projects): web-change-brief stage 4 * feat(projects): workflow-hooks stage 1 * feat(projects): workflow-hooks stage 2 * feat(projects): workflow-hooks stage 3 * feat(projects): workflow-hooks stage 4 * feat(projects): publish the practical application catalog * feat(projects): add 52 planned builds to the roadmap * fix(projects): make lesson animations explain computed changes Stage mechanisms should expose the decisions learners are implementing. Preserve record identity as state changes and keep project context below the selected lesson. * fix(projects): agent-budget-planner stage 01 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): agent-budget-planner stage 02 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): agent-budget-planner stage 03 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): agent-budget-planner stage 04 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): agent-trace-debugger stage 03 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): browser-agent stage 02 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): browser-agent stage 03 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): browser-agent stage 04 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): calendar-focus-planner stage 01 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): cloud-agent-with-aws-strands stage 01 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): cloud-agent-with-aws-strands stage 02 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): cloud-agent-with-aws-strands stage 04 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): desktop-control stage 04 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): distributed-eval-farm stage 01 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): distributed-eval-farm stage 02 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): distributed-eval-farm stage 04 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): document-extraction-desk stage 03 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): durable-agent-jobs stage 01 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): durable-agent-jobs stage 02 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): durable-agent-jobs stage 03 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): durable-agent-jobs stage 04 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): feedback-theme-board stage 01 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): inbox-triage-desk stage 03 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): llm-gateway-with-fallbacks stage 01 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): llm-gateway-with-fallbacks stage 04 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): mcp-at-scale review corrections Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): meeting-notes-to-actions stage 01 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): meeting-notes-to-actions stage 04 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): memory-server stage 04 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): pr-review-reporter stage 01 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): pr-review-reporter stage 03 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): pr-review-reporter stage 04 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): research-report-agent stage 01 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): retrieval-evaluation-lab review corrections Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): rust-agent-shell stage 01 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): rust-agent-shell stage 04 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): semantic-notes-search stage 03 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): skill-installer review corrections Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): skill-router stage 02 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): skill-router stage 04 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): tiny-coding-agent stage 02 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): voice-note-transcriber-pipeline stage 03 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): web-change-brief stage 03 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): workflow-hooks stage 01 contracts Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * fix(projects): site review corrections Keep the stage contract, learner scaffold and verification evidence consistent at the reviewed failure boundary. * docs(projects): record verified review corrections
Tiny Coding Agent
A repair agent that exposes the patch/test evidence trail and refuses unsupported success claims.
The Orchard basket repair runs in a disposable copy of a trusted Python workspace. Resolve every requested file under that root before opening it. A path may look local while a symlink redirects it outside the workspace.
Start with a learner workspace
The agent loop, Data management. Language foundation: Python data structures.
Use Python 3.10+ and its standard library. POSIX is required where the project uses file locks or process-group supervision.
python3 scripts/project_test.py tiny-coding-agent --init learning-artifacts/tiny-coding-agent
python3 scripts/project_test.py tiny-coding-agent --stage 1 --path learning-artifacts/tiny-coding-agent
Build route
- Confine file tools to a workspace
- Apply exact patches with preconditions
- Run real tests through one allowed command
- Stop the coding loop on evidence or budget
Run with your own inputs
After completing the stages, these commands run your workspace code on the original Orchard examples. Replace the sample paths with your own files.
python3 learning-artifacts/tiny-coding-agent/cli.py projects/tiny-coding-agent/examples/workspace projects/tiny-coding-agent/examples/proposals.json --copy-to basket-repair --out repair-trace.json
To inspect the complete reference first, replace learning-artifacts/tiny-coding-agent with projects/tiny-coding-agent/solution in the same command. JSON results use schema_version: 1; paths and argument examples are explicit so another tool can consume them.
Integration boundary
The CLI copies a trusted Python workspace to a new destination, then runs its tests as local code. The supplied proposal planner selects prewritten repairs from observed failures; it does not invent patches. POSIX process-group timeout stops descendants, but memory and output quotas require an OS-level runner.
python3 scripts/project_test.py tiny-coding-agent --all --solution --strict
python3 scripts/project_test.py tiny-coding-agent --all --path learning-artifacts/tiny-coding-agent --strict
The first command checks the reference. The second checks your implementation. The published examples and tests are regression evidence, not a production certification or an unseen benchmark.