siclaw AGENTS.md

A project guide for Siclaw, an AI assistant for diagnosing Kubernetes systems, with terminal, local development, and production deployment modes.

In plain words
What is it for?
Use it when modifying or checking the agent core, runtime modes, skill bundles, Docker or Helm files, Kubernetes behavior, or related tests.
Why use it?
It records safety boundaries and verification commands for changes that can affect tools, user isolation, databases, containers, or Kubernetes behavior.

Instructions file for CodexOpenCode

Install

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.

agentmods
npx agentmods add instructions/scitix/siclaw/agents-md
Clone the repo
git clone --depth 1 https://github.com/scitix/siclaw

Made for: Codex, OpenCode.

Per session 829 This file is loaded in full into every session.
When invoked 829 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00829 $0.00829
Opus 5 $0.00415 $0.00415
Sonnet 5 $0.00166 $0.00166
Haiku 4.5 $0.00083 $0.00083

Measured 2d ago against content hash 163319d74622, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

siclaw AGENTS.md 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 2d 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.

AGENTS.md · 96 lines

How it starts

The opening of the file, as written. The whole thing — 96 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Siclaw Project Instructions

This file is the project-level Codex guide for siclaw_dp_cleanup. Keep it limited to stable Siclaw facts and constraints. General collaboration style belongs in ~/.codex/AGENTS.md; CLAUDE.md is the larger legacy reference when deeper background is needed.

Project Shape

Siclaw is an AI-powered SRE copilot for Kubernetes diagnostics.

One agent core supports three runtime shapes:

TUI: single-user terminal, optionally paired with local Portal
Gateway + LocalSpawner: local multi-user dev, one process/shared filesystem
Gateway + K8sSpawner: production, one isolated pod per user

Before Risky Edits

  • Preserve user changes and unrelated dirty work.
  • For security, tool execution, database, Portal snapshot, guards, memory, Docker, Helm, or K8s behavior, read the nearby code plus relevant docs/design/* or CLAUDE.md before editing.
  • Ask before editing src/core/prompt.ts; it affects core agent behavior and product tone.
  • Do not add production dependencies unless explicitly requested.

Verification

Use targeted checks first, then broaden when touching shared behavior:

npm test
npx tsc --noEmit
npm run build

Broaden verification for database schema, tool execution, guards, Portal snapshot, K8s/Docker/Helm, or user-facing workflows.

Stable Boundaries

  • LocalSpawner runs all local AgentBox instances in one process with a shared filesystem. Local skill sync must stay user-scoped and must not wipe shared skill directories such as global/skillset/user trees.
  • Core skills are baked into the Docker image. Workspace skill bundles should include only selected global/dev/personal skills.
  • TUI plus local Portal uses Portal as a read-only snapshot source. TUI startup must tolerate missing or unauthorized Portal snapshot access.
  • Shell execution security is layered: OS-level isolation first, whitelist-only command validation second, plus pre/post execution sanitization.
  • Portal/Gateway DB and Memory DB are separate persistence domains. Do not mix user/session config with embedding/chunk/investigation storage.
  • AgentBox and Runtime are separate processes in K8s mode. Code under src/agentbox/** must not import Gateway or Portal persistence modules such as src/gateway/chat-repo.ts; LocalSpawner's shared process can hide this. AgentBox background work should call Runtime internal APIs through GatewayClient, with Runtime owning Portal RPC and database persistence.
  • src/portal/migrate.ts must stay compatible with both MySQL and SQLite.
  • mTLS is for K8s mode only; do not pull mTLS requirements into local mode.

Read the full file on GitHub · 96 lines

Changes

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.

  1. 2d ago First seen · 96 lines · 829 tokens per session scan A 163319d74622

Subscribe to this mod's changes

siclaw AGENTS.md is an instructions file published in the GitHub repository scitix/siclaw (230 stars, last pushed 3d ago), licensed Apache-2.0. It adds 829 tokens to every session, about $0.0041 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.