Find the shortest path from an invalid state to a valid one. Given data that fails validation, computes a sequence of minimal field changes that would bring the data into compliance with the Blueprint's rules and constraints. Returns the path as an ordered list of intermediate states. Different f...
Risk signalsHandles credentials or secrets (api_key)
Part of the Governance Platform server.
Free to start. No card required.
AI agents call repair_path to retrieve information from Governance Platform without modifying any data. This is common in research, monitoring, and reporting workflows where the agent needs context before taking action. Because read operations don't change state, they are generally safe to allow without restrictions -- but you may still want rate limits to control API costs.
Even though repair_path only reads data, uncontrolled read access can leak sensitive information or rack up API costs. An agent caught in a retry loop could make thousands of calls per minute. A rate limit gives you a safety net without blocking legitimate use.
Read-only tools are safe to allow by default. No rate limit needed unless you want to control costs.
{
"version": "1",
"default": "deny",
"tools": {
"repair_path": {}
}
}See the full Governance Platform policy for all 31 tools.
These attack patterns abuse exactly the kind of access repair_path gives an agent. Each links to the full case and the policy that stops it:
Other read tools across the catalogue. The same approach applies to each: allow, with a rate cap to control cost.
Find the shortest path from an invalid state to a valid one. Given data that fails validation, computes a sequence of minimal field changes that would bring the data into compliance with the Blueprint's rules and constraints. Returns the path as an ordered list of intermediate states. Different from repair: repair returns the single nearest valid point on the constraint manifold (a one-shot projection). repair_path returns the trajectory between invalid and valid states — useful when you want to see what's changing step by step, or when there are multiple ways to repair the data and you want to evaluate alternatives. Use repair when you just need the fix; use repair_path when you need to explain or audit the fix. Args: api_key: GeodesicAI API key (starts with gai_) structured_data: Current (invalid) data state blueprint: Blueprint defining the valid constraint space. Caller must own the Blueprint. max_depth: Maximum repair steps to search (1-10). Higher values find paths through more complex repairs but take longer. rank_by: Ranking criterion for selecting among valid paths. One of: "shortest" — fewest changes "drift" — most stable trajectory "confidence" — highest geometric confidence "risk" — lowest risk Returns: path: ordered list of intermediate states, each with field values and structural quality metrics path_length: number of steps in the returned path terminal_state: the final valid state at the end of the path rank_score: score of the chosen path under rank_by alternatives: summary of paths considered but not chosen. It is categorised as a Read tool in the Governance Platform MCP Server, which means it retrieves data without modifying state.
Register the Governance Platform MCP server in PolicyLayer and add a rule for repair_path: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Governance Platform. Nothing to install.
repair_path is a Read tool with low risk. Read-only tools are generally safe to allow by default.
Yes. Add a rate_limit block to the repair_path rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for repair_path. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
repair_path is provided by the Governance Platform MCP server (https://app.geodesiclabs.ai/mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Deterministic rules across all 31 Governance Platform tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.
Free to start. No card required.
4,600+ MCP servers and 31,000+ tools scanned and risk-classified.