52cd0a3cb6
Three additions to agent-repos.md based on the post-A1-incident dogfood batch (8 successes + 1 operator-induced "failure"): 1. Airouter Qwen3.6 section: pattern reconfirmed across M16 Wave A1/A2/B1 and M25 Waves A1-A5 + B1-B2. Time-to-success bands recorded for cost calibration (1m30s for git rm; ~5 min for Pydantic regex; ~12 min for class addition). Default --max-test-iterations 1 for cheap probes. 2. New section: Test Design for AI Agent Dogfood Pipelines. Triggered by the M16 Wave B2 (MN-4 prompt cap) failure — a 14-minute airouter run blamed on the agent that was actually an over-strict test asserting on sanitised 422 body content. CP's RequestValidationError handler strips Pydantic detail for security; tests asserting body content for that path are structurally impossible. Rules: verify test passes against a reference impl before pushing; status-code-only ceiling for validator- driven 422s; model on previous successes not stricter variants; F70 retries don't recover structurally impossible tests. 3. New section: Dogfood Failure Path: Branch + Logs Lost. When all F70 retries exhaust, the agent's last attempt is not pushed to the agents fork, the CP task record's logs field is empty, and the pod is gone. Operator must reproduce locally — until F70 finalize-on-failure pushes the failed branch. BESTPRACTICES.md index updated to reflect the new sub-topics. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
best-practices
Cross-project best practices extracted from real project work via the /distill-best-practices skill.
How It Works
The knowledge distillation pipeline in claude-foundations processes session logs and memory files from all tracked projects, extracting generalisable practices into topic files here.
Pipeline
/log— Captures session decisions and gotchas into per-projectmemory/log//reflect-logs— Processes logs into structured topic memory files/distill-best-practices— Reads memory files across projects, proposes updates to this repo
For Humans
Browse BESTPRACTICES.md for the full index. Each topic file is self-contained.
For Agents
Container agents get this repo cloned to /best-practices. Read BESTPRACTICES.md for the index, then read only the topic files relevant to your task.
Topics
| File | Description |
|---|---|
ansible.md |
Inventory, templates, idempotency, credential safety |
database-selection.md |
SQLite vs PostgreSQL decision criteria |
debugging.md |
Systematic diagnosis, full-chain testing, common pitfalls |
docker.md |
gosu PID 1, GIT_SSH_COMMAND scope, slim image patterns |
docker-uid-matching.md |
UID wrapper entrypoint, gosu pattern |
documentation.md |
CLAUDE.md, MEMORY.md, FUTURE.md, README.md structure |
git-source-control.md |
Commit practices, GitOps workflows, remote conventions |
helm.md |
Schema validation, version verification, values structure |
kubernetes.md |
Volume mounts, deployment strategies, naming, bootstrap ordering |
linting.md |
Tool choices per language, PostToolUse hook, pre-commit |
milestones.md |
Milestone workflow, verification, reflection process |
networking.md |
nftables, systemd sockets, Docker forwarding, TLS |
octopus-process-templates.md |
OCL syntax, step templates, Platform Hub patterns |
scripting.md |
Shell conventions, verification scripts, idempotency |
secrets-management.md |
SOPS + age, credential handling, encryption gotchas |
security-architecture.md |
Server boundary rule, proxy patterns, defense in depth |
skills-development.md |
Skill authoring, context injection, tool restrictions |
spec-driven-development.md |
Spec structure, requirement numbering, test-first workflow |
test-driven-development.md |
Edge case discovery, property-based testing, AI agent patterns |
validation.md |
Validate locally, deploy once; full-chain testing |
Source Control
- Gitea:
skynet/best-practices - Remote:
git@gitea.oreillyit.nz-ai-enablement:skynet/best-practices.git