Skip to content
Back to skills

Init

ASecurity

Scan the current working directory and write project anchors so future turns load that context automatically.

  • 10 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added September 3, 2026
ai-agentsrustgobashapibackend

Works with

  • api
  • mcp

Security analysis

A100/100

Scanned September 3, 2026

npx -y skills add xgre1/troth --skill init --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Init?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Init
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/xgre1-init/badge)](https://www.skillsdirectory.com/skills/xgre1-init)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: init
description: Scan the current working directory and write project anchors so future turns load that context automatically.
allowed-tools: [Read, Glob, Grep, engram_record]
auto-persist: {"scope":"project_anchor","salience":1.5}
---

User is initializing substrate for this project. Goal: write 3–7 project-anchor engrams so subsequent sessions in this cwd load the project context for free.

Init protocol:

1. Inspect the project root:
   - `Glob({ pattern: "*.md", path: ".", limit: 5 })` — get top-level docs.
   - `Glob({ pattern: "package.json" }) || Glob({ pattern: "Cargo.toml" }) || Glob({ pattern: "pyproject.toml" })` — detect the language / build system.
   - `Read({ file_path: "<root-readme-or-claude.md>" , limit: 80 })` if one exists.
2. Synthesize 3–7 short anchors. Examples:
   - "Project is a Rust Tauri voice app; entry: src-tauri/src/lib.rs."
   - "Tests live under tests/test-all.js, runner: `npm test`."
   - "Brain backend selectable in Settings: subscription / api_key / troth_proxy / troth_entity."
3. Persist each as an engram via `engram_record({ statement, salience: 1.5, scope: "project_anchor" })`. Salience above default so they ride higher in retrieval.
4. Reply with the list of anchors persisted + the recommendation to run `/init` again whenever the project shape changes significantly.

Why this matters: every new session that opens this cwd will pull `scope:"project_anchor"` engrams into the identity envelope automatically. The agent will "know" the project without re-discovering it.

## Tool routing (both topologies)
The substrate tools this skill uses may be DIRECT in your tool list (names like
`troth_engram_record`, `troth_recall`, `troth_dialogue_recent`) OR behind the
troth-router gateway (app installs wire only: troth-router, troth-bash,
troth-cache, troth-hashline). If a named tool is NOT in your tool list, do NOT
conclude the substrate is down and do NOT fall back to file-based memory.
Route it through the router instead:
  1. `mcp_list({server: "troth-substrate"})` (or `"troth-memory"`) to see names.
  2. `mcp_call({server: "troth-substrate", tool: "<same troth_* name>", args: {...same args...}})`.
Substrate lives on server `troth-substrate` (engrams, recall, dialogue, slash);
mind/actions on `troth-memory`. Same tools, same args, one hop through mcp_call.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.