Research Rust crate candidates and their feature, API, dependency, license, build, and runtime paths for any project. Use before dependency alignment when a unit may introduce a Rust crate, use a lower backend crate, replace a capability, or hand-write behavior that an existing crate might cover.
Scanned 9/10/2026
Install to Claude Code
npx -y skills add AntheaLaffy/py2rs-loop-lab --skill crate-recon --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Crate Recon?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/anthealaffy-crate-recon)More formats (shields.io, HTML) on the badges page.
---
name: crate-recon
description: Research Rust crate candidates and their feature, API, dependency, license, build, and runtime paths for any project. Use before dependency alignment when a unit may introduce a Rust crate, use a lower backend crate, replace a capability, or hand-write behavior that an existing crate might cover.
---
# Crate Recon
Use this skill to produce compact evidence, not a dependency decision. The
consumer is usually `dependency-alignment`.
## Evidence Ladder
1. Search crates by capability, not package-name parity.
2. Include user-named candidates even if they are not in the top search results.
3. Check current docs and feature flags from primary sources when network access
is available.
4. Use Cargo metadata/tree/info to prove resolved versions and dependency paths.
5. Record license, build requirements, native/runtime risk, maintenance signals,
and API fit.
Do not reject an umbrella crate before checking whether a feature-selected lower
dependency owns the useful capability.
## Script
Use the bundled script when appropriate:
```bash
python skills/foundations/crate-recon/scripts/crate_recon.py <capability>
```
If network or registry access is unavailable, record that explicitly and return
the best local/manual evidence.
## Report Shape
```yaml
capability: "what the project needs"
mode: agent # agent | manual | disabled | blocked
candidates:
- crate: name
version: null
fit: direct | backend | partial | poor | unknown
useful_features: []
dependency_paths: []
api_evidence: []
license: null
build_runtime_risk: []
maintenance_notes: []
recommendation_input:
viable_direct: []
viable_backend: []
unresolved_questions: []
do_not_use: []
```
## Non-Negotiables
- Recon does not add dependencies or edit lockfiles.
- Recon does not decide adapter vs handwritten replacement; alignment does.
- Disabled recon is recorded as `user_disabled`, not as completed evidence.
- `/tmp` research is disposable until promoted into a project-controlled record.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!