Paste a public GitHub URL. SAICA-KG inspects the repo, identifies the supervision tools already in your stack, scores coverage across the 11 failure modes, and recommends concrete tools to close the gaps.
Public repositories only (MVP). Audit is read-only — no fork, no PR, no OAuth. The full report is also downloadable as Markdown so you can paste it into your team's docs.
Try a well-known repo
Couldn't run the audit
The backend is a separate FastAPI service. Start it withpython -m pipeline.webapp and try again.
Audit summary
Detected stack
Languages
Package mgrs
CI providers
Runtime hints
Coding agents
Supervision tools detected
Detected but not in the KG
Coverage heatmap
Rows are the supervision tools we detected in your repo. Columns are the 11 failure modes from SAICA-KG. Cell color uses the same per-mapping evidence tier as/tool-coverage: 0 (not addressed), 1 (declared), 2 (rationale discusses FM), 3 (external citation).
No supervision tools detected in this repo. Every column will appear in the gap list below.
Tool
Source
Ranked gaps
Failure modes your stack does not yet cover, ranked by severity. Each card lists 2–3 recommended supervision tools that fit your detected language, CI, and agent context.
No gaps found — your stack already covers every failure mode in the SAICA-KG taxonomy. (Worth double-checking: review the heatmap above for thin paradigm coverage.)