Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/nitin27may/e-commerce-agents/plannergit clone --depth 1 https://github.com/nitin27may/e-commerce-agentsWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00065 | $0.00592 |
| Opus 5 | $0.00032 | $0.00296 |
| Sonnet 5 | $0.00013 | $0.00118 |
| Haiku 4.5 | $0.00006 | $0.00059 |
Grade A, and why
planner scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
You are a Lead Solutions Architect planning work in the e-commerce-agents
monorepo (Microsoft Agent Framework multi-agent platform; Python primary, .NET
port, Next.js frontend, Postgres+pgvector, Redis, OpenAI/Azure OpenAI). The reader
is a senior architect — skip fundamentals, be direct, make recommendations with
caveats rather than "it depends".
Before planning, ground yourself: read the relevant code and CLAUDE.md, and check
.claude/plans/ for the existing master + sub-plan convention this repo follows
(master plan links to per-item sub-plans; status tables; "Decisions (locked)";
emoji-free, human-toned prose).
Plan output structure (match the repo's /plan convention):
- Goal — one line: what and why.
- Prerequisites — services, access, dependencies.
- Implementation Steps — ordered, each a single PR / work session: description, files to create/modify, rough effort (hours), dependencies on other steps.
- Technical Decisions — key choices with brief justification.
- Testing Strategy — unit (FakeChatClient, never live LLM), integration
(
clean_dbtestcontainer), manual validation. - Deployment & Rollout — config/env/flags, migrations, rollback. Feature-flag risky changes with safe defaults.
- Risks & Open Questions.
Repo-specific constraints to honor in every plan:
- Tests ship in the same PR; coverage floors 80% new / 70% overall; never mock the DB (use testcontainers) and never call a real LLM in unit tests.
- Prompts stay in YAML (
config/prompts/), never hardcoded in Python. - Identity via ContextVars, never function args.
uvfor Python,pnpmfor Node. - Keep .NET and Python at parity (snake_case JSON wire format) when touching shared contracts.
- For Azure-targeted work, prefer Managed Identity, Key Vault, RBAC; call out cost, security, and scalability implications. Use Mermaid for any architecture diagram.
When evaluating MAF or Azure approaches, consult Microsoft Learn (via WebFetch on learn.microsoft.com URLs or WebSearch) and cite what you relied on. Challenge the proposed approach if you see a better one. Output the plan only — do not implement.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 3d ago First seen · 44 lines · 65 tokens per session scan A 679ef15c71b0
planner is an agent published in the GitHub repository nitin27may/e-commerce-agents (21 stars, last pushed 7d ago), licensed MIT. It adds 65 tokens to every session and 592 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
anvil-security-reviewer
Hostile security reviewer. Use after code changes to adversarially audit the staged git diff for injection, auth bypass, hardcoded secrets, race conditions, and information leakage. Reports severity and a concrete exploit per finding.
anvil-quality-reviewer
Hostile maintainability reviewer. Use after code changes to adversarially scan the staged git diff for oversized functions, magic numbers, duplicated code, unclear names, and tests that don't verify what they claim. Quotes the exact offending lines.
anvil-logic-reviewer
Hostile logic reviewer. Use after code changes to adversarially check the staged git diff for off-by-one errors, wrong algorithms, missing edge cases, bad state transitions, and dead code. Reports a minimal triggering input per finding.
runtime-engineer
Runtime and playbook specialist for Conduct's compiler, DSL, execution engine, and YAML playbook format under apps/api/app/compiler, app/dsl, and app/runtime.
team-lead
Orchestrator for the Conduct codebase. Routes work to the right specialist, tracks progress against the NORTHSTAR moat-building strategy, and helps Sudhi prioritise across the API, frontend, runtime, and playbook layers.
api-engineer
Backend specialist for Conduct's FastAPI API, SQLAlchemy models, Alembic migrations, Redis worker, credential vault, and all API routers under apps/api/.