How schmouli/ai-sdlc works
AI-SDLC is a reusable AI-assisted development toolkit. Its sampled runtime branches include a Code Knowledge Graph that builds repository structure and supports retrieval and review digests, and SkillBox, whose telemetry supports learning extraction, proposed overlays, human review, and approved instruction injection. The README also documents Spec-Kit and EGM, but their directories are absent from the supplied tree, so they are not mapped here. Lifecycle-map and SDLC-pane generation are included as product capabilities. Some internal relationships are limited to README or source-index evidence.
Main entry points: cli.py SkillBox CLI · graph_builder.py Graph builder · overlays.py Overlay library · telemetry.py Event and session store
Components 30
Code Knowledge Graph
- KG commands
scripts/run_knowledge_graph.py - Source parser
src/kg/parser.py - Graph builder
src/kg/graph_builder.py - Graph model
src/kg/models.py - Lexical similarity
src/kg/similarity.py - Graph retrieval
src/kg/retrieval.py - PageRank
src/kg/pagerank.py - Co-edit links
src/kg/coedit.py - Review digest
src/kg/digest.py - KG reporting
src/kg/report.py - Retrieval evaluation
src/kg/evaluation.py
SkillBox
- SkillBox CLI
src/skillbox/cli.py - Runtime configuration
src/skillbox/config.py - Event and session store
src/skillbox/telemetry.py - Project context
src/skillbox/context.py - Skill registry
src/skillbox/registry.py - Skill composition
src/skillbox/compose.py - Learning extraction
src/skillbox/learning.py - Outcome metrics
src/skillbox/metrics.py - Overlay library
src/skillbox/overlays.py - Human review
src/skillbox/review.py - Instruction injection
src/skillbox/inject.py - SkillBox reports
src/skillbox/reports.py - MLflow sink
src/skillbox/mlflow_sink.py - Skill sweep
src/skillbox/sweep.py
SDLC Measurement
- Lifecycle map
scripts/generate_lifecycle_map.py - SDLC measurement pane
scripts/generate_sdlc_pane.py
Other components
- Developer/operator
- MLflow service
- AI development activity
Connections 44
- Developer/operator → KG commands: runs commands
README.md - KG commands → Graph builder: builds graph
scripts/run_knowledge_graph.py - KG commands → Graph retrieval: retrieves nodes
scripts/run_knowledge_graph.py - KG commands → Review digest: renders digest
scripts/run_knowledge_graph.py - KG commands → Retrieval evaluation: evaluates retrieval
scripts/run_knowledge_graph.py - Graph builder → Source parser: parses source
src/kg/graph_builder.py - Graph builder → Graph model: creates graph
src/kg/graph_builder.py - Graph builder → Co-edit links: adds co-edits
src/kg/graph_builder.py - Graph builder → Lexical similarity: uses similarity
src/kg/graph_builder.py - Graph retrieval → Lexical similarity: anchors query
src/kg/retrieval.py - Graph retrieval → PageRank: ranks results
src/kg/retrieval.py - Graph retrieval → Graph model: traverses snapshot
README.md - Review digest → PageRank: selects top files
src/kg/digest.py - Retrieval evaluation → Graph retrieval: tests retrieval
src/kg/evaluation.py - Retrieval evaluation → Lexical similarity: uses similarity
src/kg/evaluation.py - Graph builder → KG reporting: reports failures
src/kg/graph_builder.py - Developer/operator → SkillBox CLI: runs commands
README.md - SkillBox CLI → Runtime configuration: loads settings
src/skillbox/cli.py - SkillBox CLI → Event and session store: records events
src/skillbox/cli.py - SkillBox CLI → Project context: detects context
src/skillbox/cli.py - SkillBox CLI → Skill registry: lists skills
src/skillbox/cli.py - SkillBox CLI → Learning extraction: extracts learnings
src/skillbox/cli.py - SkillBox CLI → Overlay library: stages overlays
src/skillbox/cli.py - SkillBox CLI → Human review: routes decisions
src/skillbox/cli.py - SkillBox CLI → Outcome metrics: reports metrics
src/skillbox/cli.py - SkillBox CLI → SkillBox reports: renders reports
src/skillbox/cli.py - SkillBox CLI → Instruction injection: refreshes instructions
src/skillbox/cli.py - SkillBox CLI → MLflow sink: pushes telemetry
src/skillbox/cli.py - Skill composition → Event and session store: records composition
src/skillbox/compose.py - Skill composition → Skill registry: loads skill
src/skillbox/compose.py - Skill composition → Project context: uses context
src/skillbox/compose.py - Skill composition → Overlay library: applies overlays
src/skillbox/compose.py - Learning extraction → Event and session store: analyzes events
src/skillbox/learning.py - Outcome metrics → Event and session store: reads outcomes
src/skillbox/metrics.py - Learning extraction → Overlay library: informs staging
src/skillbox/learning.py - Overlay library → Human review: checks audit
src/skillbox/overlays.py - Human review → Overlay library: approves versions
src/skillbox/review.py - Instruction injection → Overlay library: loads approved versions
src/skillbox/inject.py - Instruction injection → Event and session store: reads approvals
src/skillbox/inject.py - MLflow sink → MLflow service: sends metrics
- Skill sweep → Learning extraction: schedules learning
- AI development activity → Event and session store: generates activity
README.md - AI development activity → Lifecycle map: measures lifecycle
README.md - AI development activity → SDLC measurement pane: views measurements
README.md