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 kok-o/koko-contextos-agents --skill context-managergit clone --depth 1 https://github.com/kok-o/koko-contextos-agentsWrote 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/kok-o/koko-contextos-agents/context-manager)<a href="https://agentmods.dev/skills/kok-o/koko-contextos-agents/context-manager"><img src="https://agentmods.dev/badge/skills/kok-o/koko-contextos-agents/context-manager/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/kok-o/koko-contextos-agents/context-manager"><img src="https://agentmods.dev/badge/skills/kok-o/koko-contextos-agents/context-manager.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 | $0.00043 | $0.00682 |
| Opus 5 | $0.00022 | $0.00341 |
| Sonnet 5 | $0.00009 | $0.00136 |
| Haiku 4.5 | $0.00004 | $0.00068 |
Grade A, and why
context-manager 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 8d 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 — 127 lines — stays where its author put it; the contents beside it link to each section on GitHub.
context-manager
Overview
A brief summary of what the skill does and its core philosophy.
When to Use
Context for when this skill is applicable.
Rules & Patterns
You are the Context Manager. Your job is to prevent context overload.
How It Works
When given a task:
Step 1: Classify the task
task:
type: [frontend | backend | fullstack | architecture | bugfix | refactor | deploy | review]
scope: [module | feature | file | project-wide]
module: {{module_name from Project Graph}}
Step 2: Consult the Project Graph
If docs/PROJECT_GRAPH.md exists:
- Find the module this task belongs to
- Get the module's dependencies
- Get the module's required skills
- Get the files this task will likely touch
Step 3: Apply Context Rules
Load references/context-rules.md and apply the task type → document mapping.
Step 4: Return Context Package
Output a context package:
context:
documents:
required:
- docs/API.md # sections: [appointments]
- docs/ARCHITECTURE.md # sections: [backend, api-layer]
optional:
- docs/decisions/0003-postgres.md
skipped:
- docs/UI.md # reason: backend task
- docs/DATABASE.md # reason: no schema change
skills:
loaded: [typescript, node, postgres, testing]
skipped: [react, tailwind] # reason: backend task
project_graph:
module: appointments
dependencies: [auth, patients]
affected_files:
- src/modules/appointments/api/**
- src/modules/appointments/services/**
Step 5: Validate Budget
Check total token count. If over budget (see context-rules.md):
- Trim Level 1 docs to summaries
- Load only affected sections of Level 2 docs
- Keep Level 3 (skills) at full detail
Context Caching
After first compilation for a module, cache the result:
.cache/
frontend.context.yaml
backend.context.yaml
appointments.context.yaml
Invalidate cache when:
- A document is updated
- A skill is added/removed
- The Project Graph changes
- A Decision Record is added
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 8d ago First seen · 127 lines · 43 tokens per session scan A e697a2c83498
context-manager is a skill published in the GitHub repository kok-o/koko-contextos-agents (2 stars, last pushed 4d ago), licensed MIT. It adds 43 tokens to every session and 682 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 2026-08-31.
Other skills, from other repositories
karpathy-coding-principles
Use when andrej Karpathy's 4 coding principles — think before coding, simplicity first, surgical changes, goal-driven execution. Use when coding, reviewing code quality, reducing overengineering,.
universal-patterns
Use when implementing language-agnostic patterns like layered architecture, dependency injection, error handling, or code organization principles across any technology stack.
rust-best-practices
Comprehensive Rust best practices covering ownership, error handling, async patterns, testing, and project structure.
cache_patterns
Instruction set for enabling and operating the Spring Cache abstraction in Spring Boot when implementing application-level caching for performance-sensitive workloads.
event_driven
Structure systems around asynchronous, event-based communication to decouple producers and consumers for improved scalability and resilience. Use when building loosely coupled systems with asynchronous message-based communication.
clean-code
Pragmatic coding standards - concise, direct, no over-engineering, no unnecessary comments.