applicable weight 80%
greatlordorel/agent-assistant
Language: Python
branch main
Agent-Assistant — это pet - проект, созданный для углублённого освоения разработки ИИ‑агентов с использованием языка Python. Основная цель проекта — отработать на практике ключевые подходы и паттерны, применяемые при построении современных агентных систем.
Open on SourceCraftDownload report (Markdown)snapshot 8 October 2026signal coverage 80%analysed 1 October 2026 at 21:28 UTC — not all categories measured
What the 100 is made of
category weightDocumentation 20Activity 20Security 20CI/CD 15Issues 15Code health 10
The light segment — Security: will appear with a member token, while the rest share its weight for now.
Six categories
earned / weightDocumentation and best practicessignal coverage 95%14.3–15.5 / 25
partial
Project activity18.3 / 25
scored
Securitywill appear with a member token
no data
CI/CDsignal coverage 50%0–9.4 / 18.8
partial
Issues15 / 18.8
scored
Code health12.5 / 12.5
scored
States:weight goes to the restinterval, not zero
scored
partial
no data
collection failed
Not measured: CI runswill become a group once access opens
closed by the platform
Not measured: contributing guidethe platform's window was truncated: a deep file or a recent version may have fallen outside it
couldn't see it
Strengths
what's already done- Almost no marked debt in the code
- The README introduces the project
- Commits are fresh (8 days)
- Releases keep coming (7 days)
- Tests live in the repository
What affected the score
Penalties
Documentation and best practices
- −2license filemissing
- −2dependency lockfilemissing
- −2review policyno protection
- −1branch protection policyno protection
- −1share of short commit messages — 100%
Project activity
- −2commits in the last year — 3 commits
- −2unique authors — 2 authors
- −1weekly activity — 3 weeks
- −1top author's share of commits — 67%
CI/CD
- −9SourceCraft CI configurationmissing
Issues
- −4issue resolution time — 2 days50 of 54 issues
Not measured — interval width
Not a penalty: the lower bound of the score counts the metric as zero, the upper bound as full
Documentation and best practices
- up to −1contributing guidedata not collected
CI/CD
- up to −9CI runsdata not collected
What will raise the score
by strength of impact- add a SourceCraft CI configuration (.sourcecraft/ci.yaml); the service does not see external CISourceCraft CI configurationwhy: the build and the checks are reproducible for whoever takes the code+9CI/CD
- close issues fasterissue resolution time — 2 dayswhy: an outside contributor gets an answer, and their work is not lostevidence: the repository's issue list, snapshot of 8 October 2026+4Issues
- add a license filelicense filewhy: the terms of use are stated, not guessed+2Documentation
- add a dependency lockfile+2
- set up a review policy+2
- commit more often+2
- bring in a second maintainer+2
- spread work across more weeks instead of bursts+1
- set up a branch protection policy+1
- write meaningful commit messages+1
- share the work with other contributors+1