Free to publish. One command to install.
Publish a skill, and any agent can install it.
Aback Skills is a free marketplace for agent skills. Publish one and any agent can install it over MCP, with the risk signals attached.
publish
Price to publish
Current Price to publish: $0
Checked before it goes live
Status indicators include icons and text labels for accessibility.
Structure validation
PassCommand risk analysis
PassExternal link screening
ReviewAdmin review before publishing
Pass
Why teams use it
A verified catalogue, a delivery channel that works with any MCP client, and a free tier that is genuinely usable.
Reviewed before it ships
An admin reviews every submission, and the risk signals travel with the skill.
One tool call to install
The MCP server returns the exact file to write, and the client writes it.
Free to publish
Add as many skills as you like, for free, on every plan. Always.
Platform Capabilities
A catalogue you can trust, and a channel that delivers.
Every skill is reviewed, scored and versioned, and every install goes through the same audited path.
Skill Discovery
Search by text, category or tag from the directory, the REST API or an MCP client.
Policy Validation
Scripts, links and commands are analysed before an admin ever sees the submission.
Install over MCP
Point any MCP client at the endpoint and install a skill into a project in one call.
Version Governance
Every release is a version, and check_updates tells you exactly which ones moved.
Enterprise Infrastructure
Transparent architecture from upload to verification.
The protocol pipeline is intentionally visible so technical teams can audit every stage before releasing capabilities into production agents.
Ingest -> Parse & Validate -> Analyze Risk -> Publish
Step 1 of 4
Next: Parse & Validate
Ingest
Accept markdown or archive submissions and preserve raw protocol source.
Step 2 of 4
Next: Analyze Risk
Parse & Validate
Extract instructions, workflows, examples, and schema-check the package shape.
Step 3 of 4
Next: Publish
Analyze Risk
Detect unsafe commands, external links, script tags, and execution signatures.
Step 4 of 4
Final Stage
Publish
Approve verified skills and expose transparent risk data for installation decisions.
Security And Risk
Risk analysis built into every skill lifecycle.
Aback Skills never executes uploaded content. We parse it, inspect it, score it, and an admin reviews it before it reaches the catalogue.
Trust Indicators
Signal Coverage
Links, commands, scripts
Decision Logic
Threshold + hard-block rules
Execution Policy
Static analysis only
Scan Transparency
Ingest -> Detect -> Score -> Gate
Step 1: Ingest
Packages are accepted and queued for deterministic policy scans.
Step 2: Detect
Static checks flag links, risky commands, and script execution patterns.
Step 3: Score
Signals are weighted into a visible risk score for technical review.
Step 4: Gate
Threshold and hard-reject rules determine publish readiness outcomes.
External Link Detection
100%
Every package is scanned for outbound URLs and flagged before verification.
Command Policy Coverage
42 checks
Unsafe command signatures including rm -rf, wget, curl, and sudo are evaluated.
Auto-Reject Guard
70+
Packages crossing critical risk thresholds are blocked from approval workflows.
Audit Visibility
Realtime
Security outcomes are displayed with clear rationale for faster technical review.
FAQ
Clear answers for trust, risk, and adoption decisions.
Review how the platform handles verification, governance, and rollout safety before teams install skills into production agent workflows.
Each skill package is parsed, structure-validated, and scanned for risky patterns such as unsafe commands, script tags, and external links before being approved for installation.
Build With Trust
Move from skill review to safe rollout with faster team decisions.
Publishing is free on Aback Skills, on every plan. Paid plans pay for MCP access at scale, and nothing else.
Take The Next Step
Sign Up is for new accounts. Sign In is for existing accounts. Start with Sign Up, or continue with Sign In if your account is already active.