- Home
- Skills
- Technical Documentation
- AgentLock
AgentLock
Keep coding agents aligned with project facts using concise context, bounded document discovery, safe template updates, and session handoffs.
Free
Works with the AI Agents you already use
AgentLock
Example session with this skill installed
Demonstrate AgentLock in an isolated empty demo project: preview the minimal profile, initialize it, add the web-saas profile while preserving SESSION, validate the resulting structure, and show the complete bounded API discovery receipt. Use the tools from the uploaded ZIP and distinguish Draft samples from verified project facts.
- Read your context and instructions
- Compiled the agentlock
Reproducible local CLI demonstration
This example records real execution of the tools extracted from the uploaded ZIP in an isolated empty workspace. It is a tool demonstration, not a scored agent-host or model evaluation. Paths in the receipt are displayed relative to that demo workspace.
| Operation | Observed result |
|---|---|
| Preview minimal setup | No project directory created; 14 planned changes |
| Initialize minimal profile | 14 files written |
| Add web-saas profile | 5 files written; SESSION bytes unchanged |
| Validate project structure | 0 errors, 8 warnings |
Unfilled project samples remain Draft; structural success does not establish application acceptance. Populate real implementation routes and check the project before treating any sample as fact.
Complete discovery receipt
{
"matches": [
{
"document": "docs/04-design/external-api-spec.md",
"title": "Public interface contracts",
"status": "Draft",
"reason": "topic: api",
"section": {
"title": "Public interface contracts",
"line": 6
}
}
],
"omitted": 0,
"scan": {
"candidates": 12,
"read": 12,
"bytes": 33111,
"truncated": 0,
"invalid": 0,
"mappingLimited": false,
"partial": false
},
"notice": "Discovery hints only; no match does not prove that constraints are absent. Check sources and active code."
}
Next steps
Fill the project-specific entries from implementation evidence, keep SESSION as a replaceable snapshot, and run the project's own acceptance checks. Inspect matching document sections rather than loading the entire template library.
Connects securely to your tools. The creator never sees your data.
About this skill
AgentLock — Evidence-Aware Project Documentation for Coding Agents
AgentLock gives coding agents a small project-context entry and on-demand methods for finding relevant documentation, checking implementation against accepted requirements, and continuing work across sessions.
What it includes
- A short SKILL.md that routes to setup, inspection/review, or handoff guidance.
- Chinese and English documentation templates with seven selectable profiles; minimal setup avoids copying the entire topic library into a project.
- Preview-first initialization and tracked incremental updates. Project documents, configuration, and SESSION are preserved; locally modified managed tools produce conflicts instead of being silently overwritten.
- Read-only document discovery with at most five results and 2,400 output characters by default, returning references and section pointers rather than whole documents.
- Structural validation and concise, replaceable SESSION snapshots with a default advisory 60-line budget.
Example requests
“Use AgentLock to inspect this project's documentation entrypoints.”
“Set up the minimal AgentLock profile without overwriting existing work.”
“Find the contracts relevant to this change and prepare a concise handoff.”
Requirements and scope
The full workflow requires a filesystem-capable coding agent and Node.js 18 or newer. The tools have no third-party runtime dependencies. Project facts and session state stay in the target project. AgentLock does not add an agent runtime, background service, model call, or automatic extra agent.
This is a standard multi-file Agent Skills package. Local format checks and 120 tool/package regressions passed, including relocated standalone execution in both locales. Actual activation in individual agent products and delivery of supporting files through Agensi MCP still require validation. Structural checks do not establish application correctness, model improvements, token savings, or cache hits.
License and source
MIT licensed. Source and documentation: https://github.com/yjfkpyu/AgentLock
Initial project samples remain Draft until populated and checked. Structural validation does not establish application correctness. Requires Node.js 18+ and a filesystem-capable agent; actual per-host activation and Agensi MCP supporting-file delivery remain unverified. Token savings and cache hits have not been measured.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
1 use
Times used to date
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 28 days ago
- Free with an Agensi account
Requires
Permissions
Read relevant files in the explicitly selected target project. Write AgentLock documentation, project configuration, ownership manifest, and managed documentation tools only within the authorized scope. Initializer previews by default and refuses conflicting files; updates preserve project docs and SESSION. No credentials, external accounts, or network access are required for the bundled tools.