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 arctan303/creator.skill --skill project-knowledgegit clone --depth 1 https://github.com/arctan303/creator.skillWrote 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/arctan303/creator.skill/project-knowledge)<a href="https://agentmods.dev/skills/arctan303/creator.skill/project-knowledge"><img src="https://agentmods.dev/badge/skills/arctan303/creator.skill/project-knowledge/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/arctan303/creator.skill/project-knowledge"><img src="https://agentmods.dev/badge/skills/arctan303/creator.skill/project-knowledge.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 | $0.00055 | $0.00946 |
| Opus 5.5 | $0.00022 | $0.00378 |
| Sonnet 5.5 | $0.00011 | $0.00189 |
| Haiku 4.5 | $0.00006 | $0.00095 |
Grade A, and why
project-knowledge 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 7d 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 — 54 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Project knowledge
Purpose
Keep the project's current state, effective expectations and active work easy to recover. Capture useful project conventions from ordinary collaboration without requiring a separate “remember this” request. Preserve one authoritative home for each rule and the evidence for changes.
Trigger
- The user states or corrects a project design, development or release convention during an authorized task.
- Implemented or verified work changes the relevant current-state entry, limitation or task status.
- A capability or decision is replaced or removed, affecting current documentation or unfinished work.
- The user requests repository orientation, documentation reconciliation, cleanup or migration. Orientation is read-only unless editing is also authorized.
Do not run a repository-wide sweep on every turn. Update only the affected scope at meaningful decisions, handoffs or milestones. A transient tool failure belongs in the task; a workflow defect belongs to self-evolver. New product choices still belong to the product/design skills.
Required context
- Repository instructions, working-tree changes and the user's effective scope, authorization and restrictions.
- Registered entrypoints in
.creator/docs-state.json, or verified equivalent README, MANUAL or PROJECT-CONTEXT files. - The affected current rules, relevant active tasks and evidence; historical material only when needed to resolve a replacement or source.
- The user's actual statement or a stable task/decision reference. Distinguish it from an inference and from observed implementation.
Workflow
- Identify the narrow knowledge change and its existing owner. Reuse Design-Principles, Design-Brief, Development-Release or their equivalents; do not create a universal memory file containing copies of them.
- Classify the information as an explicit project convention, a verified project fact, an inferred candidate, or a task-specific authorization/constraint. Preserve its scope and source. Clear project conventions can be captured immediately; frequency is not a prerequisite. Inferences remain candidates and do not become user decisions through repetition or silence.
- Reconcile current clauses in place, including affected references and exceptions. Record what replaces or withdraws the old clause. Retain the original historical outcome and unknowns; do not merely append an opposite rule.
- For affected tasks, decide current applicability separately from historical implementation, validation, review and release. Removed capabilities close obsolete validation obligations as retired, not passed. For partial replacement, retain the still-effective outcomes and obligations with their actual evidence.
- Update verified current facts and relevant active-plan links after implementation or verification. Keep unimplemented requirements explicit. Use a small entrypoint/index with links; keep detailed decisions, commands and evidence in their existing homes.
- At an authorized cleanup or milestone, consolidate only the affected material, preserve unique evidence and repair navigation. Do not delete a document because it is old, large or unreferenced. Do not alter concurrent work, sibling projects or global preferences without scope authorization.
- Check relevant structure and links, then review the meaning of the changes: no invented decisions, resurrected retired tasks, lost live requirements or inflated evidence. Follow the user's language and the existing document's language. Briefly report the actual update and any material uncertainty.
What ships with it
2 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.
- 7d ago First seen · 54 lines · 55 tokens per session scan A a6be56750d1d
project-knowledge is a skill published in the GitHub repository arctan303/creator.skill (19 stars, last pushed 7d ago), licensed MIT. It adds 55 tokens to every session and 946 once invoked, about $0.0002 per session on Opus 5.5. 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 2026-10-15.
Other skills, from other repositories
berkahkarya-orchestrator
Use when orchestrate multi-skill workflows by routing tasks to the right agents and coordinating cross-platform operations.
teamwork
Use when dynamically creates and manages AI agent teams for complex tasks. Invoke when user requests multi-agent collaboration, complex project execution, or when tasks require specialized roles and coordinated workflow. Use when working with teamwork.
neat-freak
Knowledge and governance closeout: reconcile project docs, rule files (CLAUDE.md/AGENTS.md), authorized agent memory, and workspace residue with what the code and runtime actually do, so the next session or the next person starts from one current answer. Trigger when the user names "neat-freak", "洁癖", or "/neat" — and…
multi-agent
A process for coordinating multiple coding agents by splitting work among scouts, workers, reviewers, and repair agents. Each agent has a defined role and task.
Self-Improving + Proactive Agent
Self-reflection + Self-criticism + Self-learning + Self-organizing memory. Agent evaluates its own work, catches mistakes, and improves permanently. Use when (1) a command, tool, API, or operation fails; (2) the user corrects you or rejects your work; (3) you realize your knowledge is outdated or incorrect; (4) you…
tencent-meeting-mcp
A Tencent Meeting assistant for scheduling meetings and working with participants, recordings, transcripts, and AI-generated meeting notes.