ai.geodesiclabs/governance-platform

Measured conclusion: Deep-tested. L2 outcome: tools_ok.

Pre-execution governance for AI agents. Deterministic PASS/FAIL/REVIEW verdicts, replayable proof.

Add to your agent

Copy this into your MCP client config:

{
  "mcpServers": {
    "governance-platform": {
      "url": "https://app.geodesiclabs.ai/mcp",
      "transport": "streamable-http"
    }
  }
}

Measured tools (37)

• validate

  Validate structured data against a Blueprint's rules BEFORE the result is used. Returns PASS, FAIL, or REVIEW with plain-language findings, repair suggestions, a determinism hash, and a re-verifiable certificate. Same input + same rules = same verdict, every time.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • structured_data (required) — object — The document's extracted fields as key/value pairs. Keys are open by design - your Blueprint's rules define what is checked
  • blueprint — string — Blueprint name (workflow_name) to use
  • source_text — Optional. The text of the document the fields were extracted from. When the Blueprint has a source check, extracted values are compared against this text. It is used for the check only and is not stored

  Hints: read-only no · destructive no · idempotent no · open-world no

• validate_repair

  Validate structured data against a Blueprint and, when it fails, include repair suggestions (corrected values with the rule each fix is based on) in the same call. Same verdicts as validate: PASS, FAIL, or REVIEW, with reasons and proof.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • structured_data (required) — object — The document's extracted fields as key/value pairs. Keys are open by design - your Blueprint's rules define what is checked
  • blueprint — string — Blueprint name (workflow_name) to use

  Hints: read-only no · destructive no · idempotent no · open-world no

• create_blueprint

  Create a Blueprint - the governance contract validation runs against. A Blueprint defines what correct means for your data: fields, the math that must hold between them, and acceptable ranges. Start from load_rule_pack or discover_patterns if you have no rules yet; invoke the blueprint_guide prompt for the full rule/constraint reference. Returns the new Blueprint's API key.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • customer_name (required) — string — Organization or project name (also used for storage folder naming)
  • workflow_name (required) — string — Unique Blueprint identifier; the value passed as 'blueprint' in validate
  • mode — string — observe: platform checks the agent's work; enforce: platform computes derived fields itself
  • require_math — boolean — Validate mathematical relationships
  • derived_fields — array — Field names the platform computes from other fields, e.g. ['subtotal','total']
  • semantic_checks — array — Domain-specific semantic check objects
  • derivation_rules — array — Math rules as objects. Types: add, subtract, multiply, divide, round, copy, sum (multi-operand), items_multiply, items_sum. Each needs 'type' plus its fields; see the blueprint_guide prompt
  • extracted_fields — array — Field names the agent extracts from source data, e.g. ['vendor','qty','unit_cost']
  • require_coherence — boolean — Check cross-field plausibility
  • formal_constraints — array — Constraint objects. Types incl. magnitude_anchor {field,min,max}, relative_anchor {field,reference_field,ratio_min,ratio_max}, max_action_threshold {field,threshold,on_violation}, required_fields {fields}, equals, range, in_set, regex_match, items_magnitude_anchor; see the blueprint_guide prompt
  • require_provenance — boolean — Require extraction source locations for fields
  • require_consistency — boolean — Check internal field consistency
  • enable_drift_tracking — boolean — Track pattern stability across batches
  • require_high_assurance — boolean — Strictest mode: every check must pass
  • enable_anomaly_detection — boolean — Flag records that break no rules but do not fit the reference pattern

  Hints: read-only no · destructive no · idempotent no · open-world no

• list_blueprints read-only

  List the Blueprints on this account with field/rule/constraint counts and mode. Use the returned workflow_name as 'blueprint' in validate.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• repair read-only

  One-shot repair: return corrected values that would make failing data valid under the Blueprint. Use repair_path to see the steps instead.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • structured_data (required) — object — The document's extracted fields as key/value pairs. Keys are open by design - your Blueprint's rules define what is checked
  • blueprint — string — Blueprint name (workflow_name) to use
  • derivation_rules — array — Math rules as objects. Types: add, subtract, multiply, divide, round, copy, sum (multi-operand), items_multiply, items_sum. Each needs 'type' plus its fields; see the blueprint_guide prompt
  • formal_constraints — array — Constraint objects. Types incl. magnitude_anchor {field,min,max}, relative_anchor {field,reference_field,ratio_min,ratio_max}, max_action_threshold {field,threshold,on_violation}, required_fields {fields}, equals, range, in_set, regex_match, items_magnitude_anchor; see the blueprint_guide prompt

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• check_blueprint_health read-only

  Static pre-deploy analysis of a Blueprint's rule set. Returns a health verdict - healthy, acceptable, fragile, rigid, split, brittle_islands, or unsatisfiable - with advice, including joint conflicts pairwise checks miss.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • config — object — Raw blueprint config with derivation_rules and formal_constraints (used when 'blueprint' is not given)
  • blueprint — string — Blueprint name (workflow_name) to use

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• compare_semantic_equivalence read-only

  Compare two payloads under the dual-hash design: content_hash is content_hash normalizes field order and numeric formatting. Semantic comparison preserves field roles; renaming requires an explicit bijection. With supplied rules, scalar types and whitespace remain significant. Structural similarity alone does not establish decision equivalence.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • payload_a (required) — object — First structured payload (arbitrary JSON object)
  • payload_b (required) — object — Second structured payload to compare against payload_a
  • rules_a — array — Derivation rules for A
  • rules_b — array — Derivation rules for B
  • constraints_a — array — Formal constraints for A
  • constraints_b — array — Formal constraints for B
  • field_mapping — object — Explicit one-to-one field renaming from A to B

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• govern_inference

  Quality-govern an in-progress AI generation step BEFORE its output is used (complements validate, which checks finished documents). Returns an action - STOP, CONTINUE, REPAIR_REGION, REUSE_MOTIF, REVIEW, ESCALATE - with a plain-language explanation. Structural scores do not establish task correctness. Check safe_to_finalize and acceptance coverage. Persistence success is reported; read traces in the same Blueprint namespace.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • task_type (required) — string — Kind of generation step being governed
  • payload (required) — object — Task-type payload: generative_text {text,...}; retrieval {query,candidates}; generic {features}
  • inference_id (required) — string — Caller-chosen ID grouping the steps of one generation
  • source — string — Free-form caller label recorded for audit
  • blueprint — string — Owned Blueprint namespace for the trace
  • step_index — integer — Step number within this generation (0-based)
  • constraints — object — Optional governance constraint config object

  Hints: read-only no · destructive no · idempotent no · open-world no

• get_inference_trace read-only

  Retrieve the durable audit trail for a governed generation: every recorded decision and its reasons.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • inference_id (required) — string — Caller-chosen ID grouping the steps of one generation
  • blueprint — string — Blueprint namespace used when recording the trace
  • blueprint_version — string — Optional historical blueprint_version hash returned by govern_inference

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• recent_inference_decisions read-only

  Recent generation-governance decisions in this owner's Blueprint version.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • limit — integer — Maximum rows to return
  • action — string — Optional action filter (STOP, CONTINUE, REVIEW, ...)
  • blueprint — string — Blueprint namespace used when recording the trace
  • blueprint_version — string — Optional historical blueprint_version hash returned by govern_inference

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• verify_certificate read-only

  Independently re-verify a validation certificate. Integrity mode checks the hash chain; full mode (certificate + original data) recomputes every attested rule from scratch - trust nothing, recheck everything.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • certificate (required) — object — The certificate object from a prior validation response
  • data — object — Original payload for full re-verification; omit for integrity-only mode
  • derivation_rules — array — Rule list for independent attestation recomputation (optional)
  • formal_constraints — array — Optional constraints to match against the committed bundle

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• profile_blueprint_robustness read-only

  Sweep the Blueprint's numeric constraint bounds and report verdict stability: the stable band, the scales where the verdict first flips, and advice. Use before deploying bound changes.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • config — object — Raw blueprint config to profile (used when 'blueprint' is not given)
  • blueprint — string — Blueprint name (workflow_name) to use

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• forecast read-only

  Deterministic forward reasoning: from the current data state, generate and rank the valid next states reachable under the Blueprint's rules.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • structured_data (required) — object — The document's extracted fields as key/value pairs. Keys are open by design - your Blueprint's rules define what is checked
  • rank_by — string — Ranking criterion for returned paths
  • blueprint — string — Blueprint name (workflow_name) to use
  • max_depth — integer — Search depth, 1-10
  • max_branches — integer — Branches per step, 1-10

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• discover_patterns

  Learn candidate validation rules and structural document types from a batch of your records, deterministically - no Blueprint required. Promote results with approve_rule. Source data is not stored.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • documents (required) — array — List of structured records (objects) to analyze
  • blueprint — string — Blueprint name (workflow_name) to use

  Hints: read-only no · destructive no · idempotent no · open-world no

• repair_path read-only

  Find the shortest sequence of field changes taking invalid data to a valid state, as an ordered path of intermediate states. Different from repair (one-shot nearest fix): use repair_path to explain or audit the fix, or compare alternative repairs.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • structured_data (required) — object — The document's extracted fields as key/value pairs. Keys are open by design - your Blueprint's rules define what is checked
  • rank_by — string — Ranking criterion for returned paths
  • blueprint — string — Blueprint name (workflow_name) to use
  • max_depth — integer — Search depth, 1-10

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• counterfactual read-only

  Run the same data under two rule sets and compare which future states remain valid - what-if analysis for rule changes.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • structured_data (required) — object — The document's extracted fields as key/value pairs. Keys are open by design - your Blueprint's rules define what is checked
  • rules_b — array — Alternative derivation rules (rule set B)
  • blueprint — string — Blueprint name (workflow_name) to use
  • constraints_b — array — Alternative constraints (rule set B)

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• analyze_anomaly read-only

  Explain whether a record fits the usual pattern for records like it, and which fields stand out. No Blueprint required.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • structured_data (required) — object — The document's extracted fields as key/value pairs. Keys are open by design - your Blueprint's rules define what is checked
  • blueprint — string — Discovery namespace used by discover_patterns

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• create_chain

  Create a multi-agent sequential chain: stages validate in order against one Blueprint, repairs propagate forward, TTL bounds the run. Siblings: submit_chain_stage advances the chain; handoff_audit verifies a transition between stages. Returns chain_id.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • blueprint (required) — string — Blueprint governing all stages of the chain
  • stages (required) — array — Stage definitions, e.g. [{'stage_name':'extract','agent_name':'PDF Agent'}]; minimum 2
  • ttl — integer — Chain timeout in seconds; stages cannot advance after expiry

  Hints: read-only no · destructive no · idempotent no · open-world no

• submit_chain_stage

  Submit data for the chain's current stage; the platform validates it and advances the chain if it passes. Response includes next-stage info and accumulated repairs.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • chain_id (required) — string — Chain identifier returned by create_chain
  • stage (required) — string — Stage name to submit for (must be the chain's current stage)
  • structured_data (required) — object — The document's extracted fields as key/value pairs. Keys are open by design - your Blueprint's rules define what is checked

  Hints: read-only no · destructive no · idempotent no · open-world no

• handoff_audit read-only

  Audit a handoff between two chain stages: a context capsule of verified facts from the prior stage, and (if proposed_data is given) a compatibility verdict that catches fields mutated in transit. Siblings: create_chain, submit_chain_stage.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • chain_id (required) — string — Chain identifier returned by create_chain
  • from_stage (required) — string — Completed stage name (agent A)
  • to_stage (required) — string — Stage about to start (agent B)
  • proposed_data — object — Data agent B intends to submit; checked for mutation against agent A's verified fields

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• approve_rule

  Promote a rule discovered by discover_patterns into Blueprint-ready form.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • rule_id (required) — string — Discovered rule ID from discover_patterns
  • blueprint — string — Blueprint name (workflow_name) to use

  Hints: read-only no · destructive no · idempotent yes · open-world no

• reject_rule

  Reject a discovered candidate rule so it will not be promoted into a Blueprint. Pair with approve_rule after discover_patterns.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • rule_id (required) — string — Discovered rule ID from discover_patterns
  • blueprint — string — Blueprint name (workflow_name) to use

  Hints: read-only no · destructive no · idempotent yes · open-world no

• structural_types read-only

  Retrieve the document categories a discover_patterns session identified (counts, distinguishing fields, domain hints). Read-only; returns status=no_session if discovery has not run for this namespace.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • blueprint — string — Blueprint name (workflow_name) to use

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• decompose_failure read-only

  Measure the numeric difference between original and corrected values: how much lies along the supplied linear rule directions (projection) and how much lies outside them (residual), with per-field contributions. Diagnostic only: it does not establish a cause, and neither record is verified. Use with a known-correct version to diff against; use analyze_anomaly when you only have the suspicious payload. Diagnostics-tier tool.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • original_values (required) — object — Original numeric field values as {field: number}
  • corrected_values (required) — object — Corrected/expected numeric field values as {field: number}
  • blueprint — string — Load rules from this Blueprint instead of passing them inline
  • derivation_rules — array — Math rules as objects. Types: add, subtract, multiply, divide, round, copy, sum (multi-operand), items_multiply, items_sum. Each needs 'type' plus its fields; see the blueprint_guide prompt
  • formal_constraints — array — Constraint objects. Types incl. magnitude_anchor {field,min,max}, relative_anchor {field,reference_field,ratio_min,ratio_max}, max_action_threshold {field,threshold,on_violation}, required_fields {fields}, equals, range, in_set, regex_match, items_magnitude_anchor; see the blueprint_guide prompt

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• geometric_confidence read-only

  Summarize an already-computed state_vector into a confidence level (high/medium/low) with a recommendation. Post-hoc digest - use analyze_anomaly or check_drift for fresh analysis of raw data.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • state_vector (required) — object — state_vector object from a prior validate or get_execution_trace result

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• check_realization

  Structural realization analysis of a payload against the Blueprint's reference configuration (requires a 'realization' block; otherwise status=skipped). Diagnostics-tier tool; prefer validate or analyze_anomaly for standard checks.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • structured_data (required) — object — The document's extracted fields as key/value pairs. Keys are open by design - your Blueprint's rules define what is checked
  • blueprint — string — Blueprint name (workflow_name) to use

  Hints: read-only no · destructive no · idempotent no · open-world no

• check_drift

  Check whether recent submissions still match the established pattern for this Blueprint. Returns a stability verdict and observation count.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • structured_data (required) — object — The document's extracted fields as key/value pairs. Keys are open by design - your Blueprint's rules define what is checked
  • blueprint — string — Blueprint name (workflow_name) to use

  Hints: read-only no · destructive no · idempotent no · open-world no

• authorize_execution

  Go/no-go for a real-world action (payment, filing, API write): runs full validation, then the Blueprint's execution gate. authorized=true only on PASS; REVIEW means do not proceed automatically. Different from validate: validate asks is this data correct, authorize_execution asks should this action happen.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • structured_data (required) — object — The document's extracted fields as key/value pairs. Keys are open by design - your Blueprint's rules define what is checked
  • blueprint (required) — string — Blueprint name (workflow_name) to use

  Hints: read-only no · destructive no · idempotent yes · open-world no

• load_rule_pack read-only

  Load a prebuilt Blueprint template (invoices, timecards, legal, POs, claims). Call without pack_id to list packs; then create_blueprint to save a customized copy.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • pack_id — string — Rule pack ID; omit to list available packs

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• get_execution_trace

  Run validation and return the per-node execution trace (node names, deterministic flags, timing) plus the verdict and determinism hash. Use validate for normal operation; this is for debugging and audit preparation.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • structured_data (required) — object — The document's extracted fields as key/value pairs. Keys are open by design - your Blueprint's rules define what is checked
  • blueprint — string — Blueprint name (workflow_name) to use

  Hints: read-only no · destructive no · idempotent yes · open-world no

• verify_replay read-only

  Check replay commitment integrity and compare recorded execution components. Version 4 includes reference context and the final result/status. A match compares commitments; this tool does not rerun the workflow or reconstruct historical reference populations, and does not prove factual correctness.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • contract_a (required) — object — replay_contract object from one execution
  • contract_b (required) — object — replay_contract object to compare against contract_a

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• account_status read-only

  This account's plan, key usage, Blueprint counts, and the deployed platform build fingerprint (version, build, deployed).

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• delete_blueprint destructive

  Permanently delete a Blueprint and revoke its API keys. Irreversible; requires confirm=true. Account-level keys are unaffected.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • workflow_name (required) — string — Blueprint to delete; its API keys are revoked
  • confirm — boolean — Must be true to confirm this irreversible action

  Hints: read-only no · destructive yes · idempotent yes · open-world no

• update_blueprint

  Update an existing Blueprint in place. Only passed fields change; pass [] to clear a list. workflow_name cannot be renamed and existing API keys keep working. Different from create_blueprint: modifies an existing Blueprint, mints no new key.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • workflow_name (required) — string — Unique Blueprint identifier; the value passed as 'blueprint' in validate
  • mode — string — New mode: observe or enforce; omit to keep current
  • require_math — boolean — Validate mathematical relationships
  • customer_name — string — Organization or project name (also used for storage folder naming)
  • derived_fields — array — Field names the platform computes from other fields, e.g. ['subtotal','total']
  • semantic_checks — array — Domain-specific semantic check objects
  • derivation_rules — array — Math rules as objects. Types: add, subtract, multiply, divide, round, copy, sum (multi-operand), items_multiply, items_sum. Each needs 'type' plus its fields; see the blueprint_guide prompt
  • extracted_fields — array — Field names the agent extracts from source data, e.g. ['vendor','qty','unit_cost']
  • require_coherence — boolean — Check cross-field plausibility
  • formal_constraints — array — Constraint objects. Types incl. magnitude_anchor {field,min,max}, relative_anchor {field,reference_field,ratio_min,ratio_max}, max_action_threshold {field,threshold,on_violation}, required_fields {fields}, equals, range, in_set, regex_match, items_magnitude_anchor; see the blueprint_guide prompt
  • require_provenance — boolean — Require extraction source locations for fields
  • require_consistency — boolean — Check internal field consistency
  • enable_drift_tracking — boolean — Track pattern stability across batches
  • require_high_assurance — boolean — Strictest mode: every check must pass
  • enable_anomaly_detection — boolean — Flag records that break no rules but do not fit the reference pattern

  Hints: read-only no · destructive no · idempotent yes · open-world no

• list_api_keys read-only

  List this account's API keys (masked) with their Blueprint bindings.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• rotate_api_key destructive

  Replace an API key with a fresh one. The old key stops working immediately; the new key inherits its bindings.

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • key_to_rotate (required) — string — The gai_ key to rotate; it stops working immediately

  Hints: read-only no · destructive yes · idempotent no · open-world no

• delete_api_key destructive

  Permanently delete one of the caller's API keys. DESTRUCTIVE — agents using the deleted key will receive auth errors immediately. The Blueprint a key was tied to (if any) is NOT affected; only the credential is revoked. To delete a Blueprint and all its keys, use delete_blueprint. The target key can be specified two ways: - As the full key string (gai_...). - As a key_id (SHA-256 hash from list_api_keys).

  Parameters:

  • api_key (required) — string — GeodesicAI API key (gai_...)
  • key_to_delete (required) — string — The gai_ key to delete
  • confirm — boolean — Must be true to confirm this irreversible action

  Hints: read-only no · destructive yes · idempotent yes · open-world no

Measured by MCPYP probing. Untested fields are omitted, never estimated. Listing is always free; MCPYP never proxies calls or stores user tokens.