Skip to content

RNA4219/workflow-cookbook

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1,012 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Workflow Cookbook / Workflow Operations Kit

License: MIT

Language: English | 日本語 | 简体中文

Workflow Cookbook is a docs and runtime kit for workflow operations and context engineering. It bundles Birdseye/Codemap, Task Seeds, acceptance operations, reusable CI, plugin-based Evidence tracking, and a repo-agnostic adaptive improvement loop blueprint for downstream software.

Recommended read order:

  1. docs/birdseye/index.json for the lightweight node graph
  2. docs/birdseye/caps/<path>.json for focused point reads

Focus procedure:

  • Find node IDs for recently changed files within +/-2 hops from index.json
  • Read only the matching caps/*.json files

What's Included

  • Birdseye / Codemap for Markdown hub synchronization and dependency mapping
  • Operational docs centered on BLUEPRINT, RUNBOOK, EVALUATION, and CHECKLISTS
  • LLM behavior tracking through StructuredLogger plugins
  • Sample config and consumer sample for the agent-protocols Evidence contract
  • A workflow host that can connect agent-taskstate and memx-resolver as optional plugins
  • An optional, downstream-facing adaptive improvement loop blueprint for post-release reflection, recall, skill evolution, and reviewed user/workspace models
  • Reusable CI / governance workflows and validation scripts

Quick Start

  1. Read the core docs: BLUEPRINT.md, RUNBOOK.md, EVALUATION.md

  2. Refresh Birdseye:

    python tools/codemap/update.py --since --emit index+caps
  3. Record an acceptance result: docs/acceptance/README.md, docs/acceptance/ACCEPTANCE_TEMPLATE.md

  4. Review the test quality baseline: docs/addenda/J_Test_Engineering.md

  5. Check collected metrics against KPI thresholds: tools/ci/check_metrics_thresholds.py, governance/metrics_thresholds.yaml

  6. Validate Birdseye freshness: tools/ci/check_birdseye_freshness.py

  7. Try Evidence tracking: tools/protocols/README.md, examples/inference_plugins.agent_protocol.sample.json

  8. Try cross-repo plugins: tools/workflow_plugins/README.md, examples/workflow_plugins.cross_repo.sample.json

  9. Validate plugin config: tools/workflow_plugins/validate_workflow_plugin_config.py

Navigation

Skills

Common Entry Points

Birdseye / Codemap

python tools/codemap/update.py --since --emit index+caps
python tools/codemap/update.py --since --radius 1 --emit caps
python tools/codemap/update.py --targets docs/birdseye/index.json,docs/birdseye/hot.json --emit index+caps
python tools/ci/check_birdseye_freshness.py --check

Metrics Thresholds

python -m tools.perf.collect_metrics --suite qa --metrics-url <url> --log-path <path>
python tools/ci/check_metrics_thresholds.py --check --metrics-json .ga/qa-metrics.json

LLM Evidence Tracking

Task Operations

Advanced: Cross-Repo Plugins

Reusable CI

See docs/ci-config.md for downstream usage and required job semantics. Use tools/ci/check_ci_gate_matrix.py to validate gate alignment in this repo.

Supporting Docs

License

MIT. Unless noted otherwise, files copied from this repo into other projects remain under the MIT License.

About

Workflow docs, Birdseye/Codemap, Task Seeds, acceptance operations, reusable CI, context engineering, and plugin-based evidence tracking

Topics

Resources

License

Stars

Watchers

Forks

Packages

 
 
 

Contributors

Languages