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 skills add tamnguyendinh/Anvien --skill working-rulesgit clone --depth 1 https://github.com/tamnguyendinh/AnvienWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/tamnguyendinh/anvien/working-rules)<a href="https://agentmods.dev/skills/tamnguyendinh/anvien/working-rules"><img src="https://agentmods.dev/badge/skills/tamnguyendinh/anvien/working-rules/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/tamnguyendinh/anvien/working-rules"><img src="https://agentmods.dev/badge/skills/tamnguyendinh/anvien/working-rules.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What 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 5.1 | $0.00034 | $0.02251 |
| Opus 5 | $0.00017 | $0.01125 |
| Sonnet 5 | $0.00007 | $0.00450 |
| Haiku 4.5 | $0.00003 | $0.00225 |
Grade A, and why
working-rules 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 9d 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.
How it starts
The opening of the file, as written. The whole thing — 195 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Working Rules
1. Rules Are the Highest Layer of Coordination
- Before doing anything, you must read
AGENTS.mdcompletely. AGENTS.mdcoordinates how to find rules, select skills, use tools, implement, verify, and accept work.- Mandatory sequence: rules first, skills second.
- Never use summaries, prior session memory, or compacted context in place of the original raw rules.
- If context is compacted or the task changes, re-read the required original sources before taking action.
2. Every Decision Must Be Evidence-Based
- Do not deduce from verbal statements.
- Do not deduce from keywords.
- Do not deduce from filenames, module names, or terminology similarities.
- Do not project patterns from other systems, projects, or applications onto this app.
- Do not treat user statements as automatic ground truth.
- User opinions, rebuttals, or hypotheses are perspectives that require additional verification.
- Conclusions must be proven by rules, source documents, source code, runtime, or corresponding evidence.
- Do not ask the user about things already documented in the repository; independently trace the evidence first.
3. Must Understand the Entire Problem Before Acting
When the user presents an issue:
- Must identify the entire scope of the issue.
- Must read all relevant rules and documentation completely.
- Never simply find a snippet with keywords matching the user's words and jump to conclusions.
- Never shift focus to the easiest detail to fix.
- Must distinguish between root cause, symptoms, boundaries, and associated impacts.
When the user revisits a small slice:
- That slice is only a supplementary perspective.
- Never turn a small slice into the entire problem.
- Never lose sight of the original goal and scope.
4. Classify Pipelines by Real Linkages
- Each pipeline must be identified by its command, state, ownership, boundary, and outcome.
- Do not group two pipelines simply because they use similar words.
- Two pipelines are only linked when a command in one pipeline genuinely requires state from the other pipeline to make decisions.
- Do not invent business relationships just to make the architecture look more complete.
- Do not complicate an inherently simple flow unless the repository contains evidence for such complexity.
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.
- 9d ago First seen · 195 lines · 34 tokens per session scan A 7a9360c2e19f
working-rules is a skill published in the GitHub repository tamnguyendinh/Anvien (9 stars, last pushed yesterday), licensed MIT. It adds 34 tokens to every session and 2,251 once invoked, about $0.0002 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 2026-09-03.
Other skills, from other repositories
kin-retrieval
Read a codebase through Kin's semantic graph instead of grep and whole-file reads. Use when finding where something lives, what a symbol does, who calls it, or which code implements a described behavior, and when a repo has been admitted to Kin (a .kin/ directory exists).
blast-radius-review
Review a change by its blast radius, using Kin's graph to find what downstream code the change can reach. Use when reviewing a diff or pull request, deciding whether an edit is safe, or answering "what else breaks if I change this" in a repository admitted to Kin.
booboo-deploy
Stand up a Booboo brain end to end — scaffold the project, write booboo.config.yaml against a real Postgres/Supabase or JSON source, build the snapshot, then wire the REST API, the MCP server, the 3D viewer and the panel. Use when someone wants a brain built for the first time, wants to point Booboo at their own…
booboo-troubleshoot
Diagnose a Booboo brain that is not working — an empty or tiny graph, a flat starburst view, missing MCP tools, a client that cannot find the snapshot, climbing orphan counts, or a build that exits clean but produces nothing. Use when a booboo build, serve, mcp, view, panel or vault command misbehaves.
booboo
Build, query, deploy and debug a Booboo brain — one graph fusing structure, knowledge, memory, agents and automations, queryable by REST or MCP and viewable in 3D. Use when the user mentions Booboo, booboo.config.yaml, brain.json, org.booboo.json, an organigram of agents, "boot my agent from the org", a 3D system…
booboo-adapter
Feed data into a Booboo brain that the built-in postgres and json adapters do not cover — write a small config-driven adapter against the spec instead of forking the builder. Use when a source is Neo4j, an API, a CSV export, a proprietary store, or any shape the standard config cannot express.