pic-standard

PIC (Provenance & Intent Contracts) is a Python library/CLI and integration set that validates structured “action proposals” for AI agents, enforcing fail-closed checks on intent/impact, provenance trust, and evidence (hash/signature) before allowing tool execution.

Evaluated Mar 30, 2026 (21d ago)
Repo ↗ Security agentic-ai security provenance intent-contracts prompt-injection-mitigation tool-gating mcp langgraph python cli ed25519 offline
⚙ Agent Friendliness
60
/ 100
Can an agent use this?
🔒 Security
60
/ 100
Is it safe for agents?
⚡ Reliability
31
/ 100
Does it work consistently?

Score Breakdown

⚙ Agent Friendliness

MCP Quality
45
Documentation
75
Error Messages
0
Auth Simplicity
90
Rate Limits
0

🔒 Security

TLS Enforcement
60
Auth Strength
55
Scope Granularity
50
Dep. Hygiene
65
Secret Handling
70

Strengths: fail-closed action gating; explicit evidence verification options (SHA-256 hashes, Ed25519 signatures) and a trust keyring with expiry/revocation support; local-first design to avoid data leaving the machine. Uncertainties: provided content does not specify HTTP server authentication/TLS requirements or detailed key storage/secret handling; scope granularity and operational authorization beyond evidence/provenance are not fully specified.

⚡ Reliability

Uptime/SLA
0
Version Stability
55
Breaking Changes
30
Error Recovery
40
AF Security Reliability

Best When

You have agent workflows where side effects must be authorized based on verifiable provenance/evidence, and you want fail-closed gating at the tool execution boundary.

Avoid When

You cannot supply any trustworthy provenance/evidence (because PIC will likely block). Avoid where you need complex, stateful, domain-specific authorization that PIC does not model.

Use Cases

  • Guard high-impact tool calls made by LLM agents (payments, irreversible actions, sensitive data access)
  • Local/offline verification of proposed actions against evidence (SHA-256 hashes, Ed25519 signatures)
  • Tool-gating integration for agent frameworks (LangGraph, MCP) to block unsafe or unproven actions
  • Building an internal safety “action boundary” layer for agentic systems

Not For

  • Not a general-purpose policy engine for all application logic (it specifically validates PIC action proposals)
  • Not a replacement for secure tool implementations or backend authorization controls
  • Not intended to manage real payment/identity security by itself (it gates agent-side intent/evidence)

Interface

REST API
Yes
GraphQL
No
gRPC
No
MCP Server
No
SDK
Yes
Webhooks
No

Authentication

Methods: Local trust/keyring for signature verification (Ed25519) Optional HTTP bridge for /verify requests (authentication not described) Optional key resolver interface for custom trust backends
OAuth: No Scopes: No

The README describes a local-first trust keyring for verifying signature evidence (not user authentication/authorization). Authentication for the optional HTTP bridge and MCP integration is not specified in the provided content.

Pricing

Free tier: No
Requires CC: No

As an Apache-2.0 Python package, it appears to be open-source and local; no hosted pricing is mentioned.

Agent Metadata

Pagination
none
Idempotent
False
Retry Guidance
Not documented

Known Gotchas

  • PIC enforces fail-closed for high-impact proposals when trusted provenance/evidence is missing; agents may need to be adapted to produce richer Action Proposal JSON.
  • Signature evidence verification depends on having a correct trusted keyring/resolver configured; otherwise actions will be blocked.

Alternatives

Full Evaluation Report

Comprehensive deep-dive: security analysis, reliability audit, agent experience review, cost modeling, competitive positioning, and improvement roadmap for pic-standard.

AI-powered analysis · PDF + markdown · Delivered within 30 minutes

$99

Package Brief

Quick verdict, integration guide, cost projections, gotchas with workarounds, and alternatives comparison.

Delivered within 10 minutes

$3

Score Monitoring

Get alerted when this package's AF, security, or reliability scores change significantly. Stay ahead of regressions.

Continuous monitoring

$3/mo

Scores are editorial opinions as of 2026-03-30.

8642
Packages Evaluated
17761
Need Evaluation
586
Need Re-evaluation
Community Powered