ID Protocol

ID turns a person into a portable AI context that travels between tools with explicit freshness, trust, and privacy rules.

Positioning:

• ID is not another assistant. It is the protocol layer that keeps context explicit, portable, and auditable.
• It complements SET, agentsmd, and other orchestrators by making context, contracts, and hooks first-class.

Start in 5–10 minutes

Install with Homebrew:

brew install markoblogo/tap/id-protocol

1. idctl init --owner-id <owner-id>
2. fill profiles/<owner>/profile.minimal.md
3. make validate
4. make compact

Result:

• immediate AI-ready portable context
• one concrete checkpoint (profile.minimal.md)
• generated artifact (context.compact.json)

Quick path

• Lite path: minimal setup, get started fast → docs/LITE.md
• Share path: multi-tool portability → docs/SHARE.md
• Bench path: evidence and claims → docs/BENCH.md

Comparative note

• Copilot, Cursor, and Continue generate local prompt memories.
• ID gives a stable externalized profile and contract that those tools can consume.

Live Public Metrics

Runs analyzed: 4

Profile freshness score (owner markoblogo): 0.0

Key artifacts:
- profiles/markoblogo/profile.core.md: score=0.0 age=67 ttl=14
- profiles/markoblogo/profile.extended.md: score=0.0 age=66 ttl=30

Orchestration status

ID now has a real execution role inside the wider ABVX toolchain:

• SET can orchestrate repo-local ID hooks in compatible repositories
• current CI-safe hooks are pre_task and weekly_review
• ID stays the context/protocol layer, while SET stays the orchestration layer
• agentsgen remains the repo-docs/runtime companion, not a replacement for portable human context

Practical consequence:

• if a repo is ID-compatible, SET can validate or refresh human-context boundaries before and after the main repo workflow
• if a repo only needs repo-scoped agent docs, agentsgen is enough on its own

Why this beats ad-hoc prompts or chat memory in some workflows:

• system prompts are fragile and usually copied by hand across tools;
• chat-native memory is siloed inside one product and hard to audit;
• project instructions help per repo, but not across roles like writing, research, or multi-tool orchestration;
• ID makes preferences, constraints, freshness, privacy, and portability explicit and versioned.

Evidence surface:

• docs/PROOF.md (claim pack and caveats)
• benchmarks/runs/public-metrics.md (live utility summary)
• docs/MEASUREMENT.md and docs/EVALUATOR_PROTOCOL.md (method)
• docs/EVIDENCE_POLICY.md (evidence rules)

Ecosystem map:

• docs/ECOSYSTEM.md
• docs/INTEGRATIONS.md
• docs entrypoint index: docs/README.md

Release/install:

• docs/RELEASES.md

Choose One Path

1. ID Lite

Use this if you want the smallest practical entrypoint.

You get:

• a starter profile

... View full README on GitHub