Evidence Gate MCP server
io.github.Dailyaiagents/evidence-gate·v0.1.1Local checks for artifacts, declared claim evidence, and citation containment.
0 stars26 downloads/wk
Reviews
Write oneNobody has reviewed Evidence Gate yet.
If you have run it, two minutes of your experience saves the next person an afternoon.
Evidence Gate tools (4)
write = sends, deletes, buys or postsRead from the package source without running it. The installed server may list more.
audit_citationsCheck whether quoted text is literally contained in a rooted local source.
audit_claimsCheck whether claims declare available local evidence; URLs remain unverified.
summarize_verificationAggregate evidence receipts without upgrading uncertain states.
verify_artifactCheck that one rooted artifact exists, is non-empty, and obeys declared text rules.
Public scan report
scanner v0.1.10 · 2026-10-30 · same rubric, same numbers if you re-run it
- Code scan5 source files scanned25/25
- –Live reliabilityno gateway calls yet and no remote to proben/a
- –Tool poisoningtools not inspected (local package is not executed); not countedn/a
- Auth qualitylocal package, no credentials required12/15
- Maintenancelast push 7 days ago15/15
- Maintainer identityregistry namespace matches repository owner7/10
How Evidence Gate compares
Against every graded MCP server, from our own scans and uptime checks. Updated after each scan run.
- Safety score
- 91/100
- higher than 85% of 48,243 servers
- Tools
- 4
- the typical server has 7
- Tools that change things
- 0%
- typical: 0%. Fewer write tools means less an agent can break
- Downloads a week
- 26
- more than 13% of 12,306 servers with a package
What its tools let an agent do: 4 read
Install Evidence Gate in Claude Code, Cursor or VS Code
claude mcp add evidence-gate -- uvx dailyaiagents-evidence-gate
What the publisher says
From the Evidence Gate repository's README, as published. We do not edit it. Read it on GitHub
Daily AI Agent Toolkit
Local, deterministic tools for checking AI-generated work before it is released.
The toolkit contains two Model Context Protocol (MCP) servers and six portable Agent Skills. They inspect retained artifacts, declared claim support, literal citations, completion contracts, blockers, and release receipts without calling a model, uploading files, or requiring a Daily AI Agents account.
Release status: 0.1.1 patch release candidate. PyPI 0.1.0 is public; the patch corrects the case-sensitive MCP Registry ownership marker.
Why this exists
AI systems can produce convincing output without proving that required artifacts exist, cited text is present, or a completion contract was actually satisfied. These tools make those narrow checks repeatable and preserve uncertainty instead of silently upgrading it to success.
They deliberately do not determine broad truth, judge semantic entailment, approve a release, validate a live service, or establish legal or security compliance.
Included tools
Evidence Gate
- verify_artifact
- audit_claims
- audit_citations
- summarize_verification
Release Gate
- check_contract
- evaluate_completion
- format_blockers
- buildreleasereceipt
Agent Skills
- artifact-verifier
- claim-truth-gate
- completion-contract
- contract-checker
- fail-loud
- verification-bench
Install and run
Python 3.11 or newer is required.
python -m pip install dailyaiagents-evidence-gate==0.1.1
python -m pip install dailyaiagents-release-gate==0.1.1
dailyai-evidence-gate --root /absolute/path/to/workspace
dailyai-release-gate --root /absolute/path/to/workspaceUntil the packages are public, install from a clean checkout:
python -m pip install ./servers/evidence-gate ./servers/release-gateBoth servers use MCP stdio transport. A representative client configuration is:
{
"mcpServers": {
"dailyai-evidence-gate": {
"command": "dailyai-evidence-gate",
"args": ["--root", "/absolute/path/to/workspace"]
},
"dailyai-release-gate": {
"command": "dailyai-release-gate",
"args": ["--root", "/absolute/path/to/workspace"]
}
}
}See client configuration for the supported transport boundary.
Reproduce the proof
Run the complete local gate:
bash scripts/test-all.shThe release candidate includes:
- unit tests for both servers;
- deterministic self-tests for all six skills;
- 20 declared-outcome examples spanning all eight MCP tools;
- rooted-path and symlink-escape protections;
- clean-wheel installation and stdio tool-discovery checks;
- package-specific checksums, SBOMs, and build provenance.
See the examples, technical report, receipt schemas, release and recovery process, and local verification record. A passing local check proves only the condition and scope named in its receipt.
Architecture and security boundary
Live repository controls and current fail-closed publication gates are recorded in Repository controls.
MCP client
| stdio + structured inputs
v
Evidence Gate / Release Gate
| canonical path resolution beneath --root
v
Local retained files ---> structured status + scoped receiptShortened. The full README is on GitHub.
Nothing above is checked by us. What we check is on the safety report.
Evidence Gate: common questions
- Is Evidence Gate MCP server safe?
- Yes, by our scan: it is graded A (91/100). Read the Evidence Gate safety report
- How do I install Evidence Gate?
- It runs on your machine. Copy the Claude Code, Cursor, VS Code or Claude Desktop config from the install section.
- Does Evidence Gate need an API key?
- Not as far as the registry entry and our scan can tell: no credentials are declared or required.
- Is Evidence Gate maintained?
- The last commit was 13 days ago (2026-10-23). The latest release is v0.1.1.
- How does Evidence Gate compare with other MCP servers?
- safety score 91/100 (higher than 85% of 48,243 servers); tools 4 (the typical server has 7); tools that change things 0% (typical: 0%. Fewer write tools means less an agent can break). Full comparison.