Files
info-tech-canon/CODING_AGENT_BOOTSTRAP.md

357 lines
6.3 KiB
Markdown
Executable File

# Coding Agent Bootstrap: info-tech-canon
## Mission
Establish the practical repository implementation for `info-tech-canon` based on the existing seed artifacts and the markdown-first infospace/tooling direction.
The repository is intended to become the canonical markdown-first knowledge and standards space for InfoTechCanon.
---
## Current Conceptual State
The current InfoTechCanon structure distinguishes:
```text
kernel/
InfoTechCanonCore.md
InfoTechCanonKernelMap.md
models/
information-space/
landscape/
organization/
governance/
task/
access-control/
security/
data/
devsecops/
network/
observability/
standards/
tagging/
caring/
profiles/
patterns/
mappings/
assimilation/
schemas/
views/
agent/
examples/
validation/
```
Core principle:
```text
Kernel = how the canon works.
Models = broad domain structures.
Standards = cross-cutting conventions or named analytical/design frameworks.
Profiles = concrete implementation constraints.
Patterns = reusable practical solutions.
Mappings = alignment to external bodies of knowledge.
Assimilation = structured digestion of external knowledge.
```
---
## First Implementation Goal
Create a clean repository scaffold that can host the existing Markdown seed documents and support future automation.
Do not over-engineer. Start with file layout, manifests, validation stubs, and repeatable commands.
---
## Expected First Commit
Create or verify:
```text
README.md
INTENT.md
SCOPE.md
canon.yaml
kernel/
models/
standards/
profiles/
patterns/
mappings/
assimilation/
schemas/
views/
agent/
examples/
validation/
```
Add placeholder `README.md` files in empty directories so Git tracks structure.
---
## File Placement Rules
Use this placement:
```text
kernel/InfoTechCanonCore.md
kernel/InfoTechCanonKernelMap.md
models/information-space/InfoTechCanonInformationSpaceModel.md
models/landscape/InfoTechCanonLandscapeModel.md
models/organization/InfoTechCanonOrganizationModel.md
models/governance/InfoTechCanonGovernanceModel.md
models/task/InfoTechCanonTaskModel.md
models/access-control/InfoTechCanonAccessControlModel.md
models/security/InfoTechCanonSecurityModel.md
models/data/InfoTechCanonDataModel.md
models/devsecops/InfoTechCanonDevSecOpsModel.md
models/network/InfoTechCanonNetworkModel.md
models/observability/InfoTechCanonObservabilityModel.md
standards/tagging/InfoTechCanonTaggingStandard.md
standards/caring/InfoTechCanonCaringAccessGovernanceStandard.md
```
---
## Initial Automation Targets
Create minimal scripts or Make targets for:
```text
make validate
make index
make tree
make agent-briefs
```
Initial implementations may be simple stubs that check file existence and generate basic views.
Recommended early scripts:
```text
scripts/validate_structure.py
scripts/generate_views.py
scripts/check_links.py
scripts/extract_frontmatter.py
```
---
## Initial Schemas
Start with lightweight YAML/JSON schemas for:
```text
schemas/standard.schema.yaml
schemas/concept.schema.yaml
schemas/mapping.schema.yaml
schemas/profile.schema.yaml
schemas/assimilation.schema.yaml
schemas/interface-card.schema.yaml
schemas/agent-brief.schema.yaml
schemas/caring-access-descriptor.schema.yaml
```
Schemas may begin minimal. Prefer useful incremental validation over completeness.
---
## Initial Generated Views
Generate or maintain:
```text
views/kernel-overview.md
views/by-standard.md
views/by-concept.md
views/by-profile.md
views/by-mapping-target.md
views/refactoring-checklist.md
```
Generated files should include a note:
```text
<!-- generated: do not edit manually -->
```
unless they are curated manually.
---
## Important Guardrails
Do not redefine domain concepts while implementing repository tooling.
Do not collapse:
```text
Organization Role
AccessRole
CARING Canonical Role
```
Do not collapse:
```text
Actor
Subject
Principal
```
Do not collapse:
```text
Policy
Control
Evidence
```
Do not collapse:
```text
Dataset
DataStore
```
Do not collapse:
```text
Artifact
Release
Deployment
Runtime State
```
Do not collapse:
```text
Network Intent
Network Policy
Network Configuration
Observed Network State
```
Do not collapse:
```text
Alert
Incident
```
Do not treat tags as substitutes for canonical fields, relationships, policies, or evidence.
---
## CARING Placement
CARING is a specialized standard, not just an Access Control profile.
Use:
```text
standards/caring/InfoTechCanonCaringAccessGovernanceStandard.md
```
Namespace:
```text
itc-caring
```
CARING should later get:
```text
standards/caring/agent-brief.md
standards/caring/concepts/
standards/caring/profiles/
standards/caring/mappings/
standards/caring/benchmarks/
standards/caring/examples/
standards/caring/validation/
```
First CARING benchmark candidate:
```text
Kubernetes RBAC
```
---
## Recommended First Task Sequence
1. Create repository tree.
2. Add `INTENT.md`, `SCOPE.md`, and `canon.yaml`.
3. Place all seed Markdown files in the correct directories.
4. Add directory-local `README.md` placeholders.
5. Add basic validation script checking:
- required top-level files,
- required directories,
- files referenced in `canon.yaml`,
- missing agent briefs.
6. Add basic view generator for `views/by-standard.md`.
7. Add `Makefile`.
8. Add initial schemas.
9. Add CARING benchmark directory.
10. Commit as: `seed repository scaffold`.
---
## Preferred Style
Keep implementation boring and inspectable.
Use:
```text
Python for scripts
YAML for manifests and schemas
Markdown for human-facing docs
Makefile for repeatable local commands
```
Avoid introducing heavy dependencies before the basic repository structure works.
---
## Acceptance Criteria for Initial Implementation
The initial implementation is acceptable when:
```text
make validate
```
can verify:
- required files exist,
- required directories exist,
- all manifest paths resolve,
- standards/models are placed in the expected structure,
- generated views can be created or refreshed,
- missing optional artifacts are reported as warnings, not failures.
---
## Next After Scaffold
After the scaffold works:
1. Create machine-readable concept ownership index.
2. Extract concept pages from seed standards.
3. Create agent briefs per standard.
4. Create mapping skeletons.
5. Create formal CARING assimilation workspace.
6. Add link validation.
7. Add front matter validation.
8. Add retrieval/chunk manifest generation.