eval-performance — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited eval-performance (Agent Skill) and scored it 45/100 (orange). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 7 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 7 flagged
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
*.sig, SIGNATURES) outside the documentation.A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
*.sig, SIGNATURES) outside the documentation.A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
*.sig, SIGNATURES) outside the documentation.A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
*.sig, SIGNATURES) outside the documentation.A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
*.sig, SIGNATURES) outside the documentation.A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
*.sig, SIGNATURES) outside the documentation.A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
*.sig, SIGNATURES) outside the documentation.Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Evaluation is the work MSBuild does before any target runs — reading project files, processing imports, expanding globs. This skill helps you find and confirm evaluation bottlenecks. Measure first; recommend a change only when a measurement proves it is warranted.
Engage only when evaluation is measurably the bottleneck. Do NOT act when:
That is not an evaluation problem — use build-perf-diagnostics instead.
incremental-build instead.
is slow, gather one first (see below). Do not guess from reading project files.
imports, or EnableDefaultItems are only worth flagging when the numbers show they cost real time. A project that evaluates quickly needs no change.
When a pattern is present but unmeasured, report it as an observation and let the user decide — do not rewrite working configuration to match a "best practice" without evidence it costs measurable evaluation time. Prefer the smallest, most targeted change; never disable SDK defaults as a first move.
For a comprehensive overview of MSBuild's evaluation and execution model, see Build process overview.
<Import>, evaluate <PropertyGroup> top-to-bottom<ItemDefinitionGroup> metadata defaults<ItemGroup> with Include, Remove, Update, glob expansionKey insight: evaluation happens BEFORE any targets run. Slow evaluation = slow build start even when nothing needs compiling.
Use the binlog MCP server (Microsoft.AITools.BinlogMcp, exposed under the binlog MCP namespace) to analyze evaluation performance:
dotnet msbuild build.binlog -noconlog -fl "-flp:v=diag;logfile=full.log"grep -i 'Evaluation started\|Evaluation finished' full.logdotnet msbuild -pp:full.xml MyProject.csprojOnly pursue these remedies once a measurement shows item evaluation is slow and the globs are the cause; a custom glob that isn't walking large trees is fine.
**/*.cs walk the entire directory treenode_modules/, .git/, bin/, obj/ — millions of files<DefaultItemExcludes> to exclude large directoriessrc/**/*.cs instead of **/*.cs<EnableDefaultItems>false</EnableDefaultItems> only as a last resort (loses SDK defaults) — prefer the two options above first/pp output → search for <!-- Importing comments to see import treegrep 'Evaluation started.*ProjectName' full.log → if count > 1, check for differing global properties/graph)$([System.IO.File]::ReadAllText(...)) during evaluation — reads file on every evaluationdotnet msbuild -pp:full.xml~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.